Machine Learning Engineer
Machine Learning Engineer
Machine Learning Engineer
4–6 years of experience
4–6 years of experience
4–6 years of experience
4–6 years of experience
About Nutpaa
Nutpaa is an early-stage deep-tech company building real-time human movement perception and intelligence using computer vision, machine learning, and 3D modeling. Our systems run on-device, combine multi-camera and wearable capture, and are built around low-latency inference and strict privacy protection.
Our culture centers on technical depth, strong engineering ownership, peer collaboration, long-term thinking, and patent-protected innovation. This is a role for an engineer who writes excellent code and takes models all the way from research to real-time production on the edge.
Role Summary
We are looking for a Machine Learning Engineer who is, first and foremost, a strong coder someone who builds robust, efficient, production-grade ML systems, not just notebooks. You will own models across our perception and coaching stack: training and evaluation, data and feature pipelines, and the optimization and deployment that get those models running in real time on edge hardware.
Machine learning is the heavy lifting in this role but our perception stack is camera-based, so you will also do solid, mid-level computer-vision engineering: image and video processing, camera calibration, multi-view geometry, and 2D/3D pose as the inputs your models depend on. Think ML-first, with CV as a strong working competency rather than a part-time interest.
You will work hands-on alongside our computer-vision, full-stack, and (as the team grows) edge and hardware engineers. The bar for code quality is high: clean, tested, reproducible, and fast. You will own your ML track and its quality, work alongside and guide junior ML engineers on it, and follow strong SDLC and CI/CD discipline across the ML lifecycle. We are looking for someone who ships tested work and collaborates well, not a solo developer.
Key Responsibilities
1. Model Development & Training
• Build & Train: Design, train, fine-tune, and evaluate models across pose/movement understanding, sequence/temporal modeling, and multimodal or Vision-Language tasks, using PyTorch.
• Rigorous Evaluation: Define metrics, benchmarks, and error analysis; drive measurable accuracy, robustness, and reliability improvements rather than one-off results.
• Data & Features: Build data and feature pipelines dataset preparation, augmentation, synthetic data, and labeling workflows that are versioned and reproducible.
• Applied Computer Vision: Build and integrate the camera-based perception around your models image/video processing (OpenCV), camera calibration, multi-view geometry, and 2D/3D pose at a solid mid-level. ML remains the core craft; CV is the working layer that feeds it.
2. Optimization & Edge Deployment
• Model Optimization: Quantize (INT8/FP16), prune, and distill models, and manage export to ONNX, TensorRT, TFLite, or CoreML for real-time inference.
• Edge Performance: Hit strict latency, memory, and thermal budgets on Jetson-class edge and NPU targets; profile and optimize the hot paths that matter.
• Production Inference: Ship models into a real serving path reproducible builds, versioning, monitoring, and clean rollbacks.
3. Engineering Excellence (Strong Coding)
• Production-Grade Code: Write clean, efficient, well-tested Python (and performance-sensitive C++ where needed); strong grasp of data structures, algorithms, and complexity.
• Software Discipline: Unit/integration tests, code reviews, profiling, clear documentation, Git, and CI/CD for models and pipelines.
• Reproducibility: Experiment tracking, deterministic pipelines, and MLOps practices so results are reproducible and deployable, not just demoable.
4. Track Ownership, Mentoring & Teamwork
• Own Your ML Track: Take ownership of your ML track the models and components assigned to you and deliver them well: designed, tested, and cleanly integrated, working closely with the wider team.
• Work With & Guide Juniors: Work alongside junior ML engineers on your track guide their design and code reviews, help grow their skills, and unblock them.
• Test Your Work: Own the quality of what you ship. Unit and integration tests, model validation, and regression checks are part of “done” not an afterthought. We expect engineers who test their work, not just develop it.
• SDLC on the ML Side: Own the software-development lifecycle for your ML work Git branching and pull requests, code reviews, automated testing, and CI/CD for both code and models, through deployment and monitoring.
• Team Player: Collaborate openly and reliably. Partner day-to-day with CV, full-stack, and edge/hardware engineers, communicate proactively, and flag dependencies and blockers early. This is a team sport, not a solo one.
What We Are Looking For
Experience & Coding
• Background: 4–6 years building and shipping machine-learning systems in production, with a solid record of taking models from research to deployment.
• Strong Coder (must-have): Excellent, demonstrable coding ability clean, efficient, production code; solid data structures, algorithms, and complexity fundamentals. A public code profile (GitHub / competitive programming) is a strong plus.
Machine Learning
• Core ML/DL: Deep, hands-on experience with PyTorch (and/or TensorFlow); CNNs, sequence/temporal models, and the training/evaluation lifecycle end to end.
• Applied Depth: Strength in one or more of: computer vision / pose estimation, multimodal or Vision-Language models, or real-time / time-series modeling.
Computer Vision (mid-level - required)
• Working CV: Solid, hands-on proficiency with OpenCV and image/video processing.
• Applied Geometry / Pose: Practical experience with camera calibration, multi-view geometry, or 2D/3D pose estimation. Deep specialization is not required but real, hands-on CV exposure is, since it feeds the models you build.
Optimization & Deployment
• Edge & Optimization: Hands-on with quantization and model optimization, and export/runtime tools (TensorRT, ONNX Runtime, TFLite, or CoreML).
• Targets: Experience deploying real-time models to GPUs and, ideally, edge/NPU or mobile hardware.
Foundations
• Toolkit: Python (NumPy, Pandas), strong debugging and profiling, Git, Linux, and Docker.
• MLOps: Experiment tracking, model versioning, and reproducible training/serving pipelines.
Engineering Discipline & Ways of Working
• Testing & Quality (must-have): A demonstrated habit of testing your own work unit and integration tests, model validation, and CI-gated quality checks. You treat untested code as unfinished.
• SDLC & CI/CD: Solid command of the software development lifecycle: Git branching and pull requests, code reviews, and CI/CD pipelines for both code and models, through deployment and monitoring.
• Working With Juniors: Comfortable guiding junior engineers on a shared track, and delivering your own work to production as part of a team not building in isolation.
• Team Player (must-have): A genuine collaborator: communicative, reliable, and generous with help. You own delivery without working in a silo and raise blockers early.
Nice to Have
• Advanced ML: VLM fine-tuning (LoRA/QLoRA) or on-device / edge ML experience.
• Education / Recognition: Master’s or Ph.D. in a relevant field, publications (CVPR/ICCV/NeurIPS), or granted patents.
• Domain: Biomechanics, fitness/sports analytics, health-tech, or AR/VR.
What You’ll Gain
• Ownership: Own real models end to end training through real-time edge deployment on a cutting-edge human-movement platform.
• Deep-Tech Craft: Work on hard, latency-critical, privacy-first ML rather than dashboards and glue.
• Patent Creation: Direct involvement in building patentable core technology.
• Growth: A clear path to senior/lead ML as the platform and team scale.
How to Apply
Interested candidates can send a CV and a short note about relevant work ideally with links to code, papers, or shipped ML systems to careers@nutpaa.com.
Please use the subject line “Machine Learning Engineer.”
Location: Madurai, Tamil Nadu · Hybrid · Full-time.
Apply now to join us
Apply now to join us
About Nutpaa
Nutpaa is an early-stage deep-tech company building real-time human movement perception and intelligence using computer vision, machine learning, and 3D modeling. Our systems run on-device, combine multi-camera and wearable capture, and are built around low-latency inference and strict privacy protection.
Our culture centers on technical depth, strong engineering ownership, peer collaboration, long-term thinking, and patent-protected innovation. This is a role for an engineer who writes excellent code and takes models all the way from research to real-time production on the edge.
Role Summary
We are looking for a Machine Learning Engineer who is, first and foremost, a strong coder someone who builds robust, efficient, production-grade ML systems, not just notebooks. You will own models across our perception and coaching stack: training and evaluation, data and feature pipelines, and the optimization and deployment that get those models running in real time on edge hardware.
Machine learning is the heavy lifting in this role but our perception stack is camera-based, so you will also do solid, mid-level computer-vision engineering: image and video processing, camera calibration, multi-view geometry, and 2D/3D pose as the inputs your models depend on. Think ML-first, with CV as a strong working competency rather than a part-time interest.
You will work hands-on alongside our computer-vision, full-stack, and (as the team grows) edge and hardware engineers. The bar for code quality is high: clean, tested, reproducible, and fast. You will own your ML track and its quality, work alongside and guide junior ML engineers on it, and follow strong SDLC and CI/CD discipline across the ML lifecycle. We are looking for someone who ships tested work and collaborates well, not a solo developer.
Key Responsibilities
1. Model Development & Training
• Build & Train: Design, train, fine-tune, and evaluate models across pose/movement understanding, sequence/temporal modeling, and multimodal or Vision-Language tasks, using PyTorch.
• Rigorous Evaluation: Define metrics, benchmarks, and error analysis; drive measurable accuracy, robustness, and reliability improvements rather than one-off results.
• Data & Features: Build data and feature pipelines dataset preparation, augmentation, synthetic data, and labeling workflows that are versioned and reproducible.
• Applied Computer Vision: Build and integrate the camera-based perception around your models image/video processing (OpenCV), camera calibration, multi-view geometry, and 2D/3D pose at a solid mid-level. ML remains the core craft; CV is the working layer that feeds it.
2. Optimization & Edge Deployment
• Model Optimization: Quantize (INT8/FP16), prune, and distill models, and manage export to ONNX, TensorRT, TFLite, or CoreML for real-time inference.
• Edge Performance: Hit strict latency, memory, and thermal budgets on Jetson-class edge and NPU targets; profile and optimize the hot paths that matter.
• Production Inference: Ship models into a real serving path reproducible builds, versioning, monitoring, and clean rollbacks.
3. Engineering Excellence (Strong Coding)
• Production-Grade Code: Write clean, efficient, well-tested Python (and performance-sensitive C++ where needed); strong grasp of data structures, algorithms, and complexity.
• Software Discipline: Unit/integration tests, code reviews, profiling, clear documentation, Git, and CI/CD for models and pipelines.
• Reproducibility: Experiment tracking, deterministic pipelines, and MLOps practices so results are reproducible and deployable, not just demoable.
4. Track Ownership, Mentoring & Teamwork
• Own Your ML Track: Take ownership of your ML track the models and components assigned to you and deliver them well: designed, tested, and cleanly integrated, working closely with the wider team.
• Work With & Guide Juniors: Work alongside junior ML engineers on your track guide their design and code reviews, help grow their skills, and unblock them.
• Test Your Work: Own the quality of what you ship. Unit and integration tests, model validation, and regression checks are part of “done” not an afterthought. We expect engineers who test their work, not just develop it.
• SDLC on the ML Side: Own the software-development lifecycle for your ML work Git branching and pull requests, code reviews, automated testing, and CI/CD for both code and models, through deployment and monitoring.
• Team Player: Collaborate openly and reliably. Partner day-to-day with CV, full-stack, and edge/hardware engineers, communicate proactively, and flag dependencies and blockers early. This is a team sport, not a solo one.
What We Are Looking For
Experience & Coding
• Background: 4–6 years building and shipping machine-learning systems in production, with a solid record of taking models from research to deployment.
• Strong Coder (must-have): Excellent, demonstrable coding ability clean, efficient, production code; solid data structures, algorithms, and complexity fundamentals. A public code profile (GitHub / competitive programming) is a strong plus.
Machine Learning
• Core ML/DL: Deep, hands-on experience with PyTorch (and/or TensorFlow); CNNs, sequence/temporal models, and the training/evaluation lifecycle end to end.
• Applied Depth: Strength in one or more of: computer vision / pose estimation, multimodal or Vision-Language models, or real-time / time-series modeling.
Computer Vision (mid-level - required)
• Working CV: Solid, hands-on proficiency with OpenCV and image/video processing.
• Applied Geometry / Pose: Practical experience with camera calibration, multi-view geometry, or 2D/3D pose estimation. Deep specialization is not required but real, hands-on CV exposure is, since it feeds the models you build.
Optimization & Deployment
• Edge & Optimization: Hands-on with quantization and model optimization, and export/runtime tools (TensorRT, ONNX Runtime, TFLite, or CoreML).
• Targets: Experience deploying real-time models to GPUs and, ideally, edge/NPU or mobile hardware.
Foundations
• Toolkit: Python (NumPy, Pandas), strong debugging and profiling, Git, Linux, and Docker.
• MLOps: Experiment tracking, model versioning, and reproducible training/serving pipelines.
Engineering Discipline & Ways of Working
• Testing & Quality (must-have): A demonstrated habit of testing your own work unit and integration tests, model validation, and CI-gated quality checks. You treat untested code as unfinished.
• SDLC & CI/CD: Solid command of the software development lifecycle: Git branching and pull requests, code reviews, and CI/CD pipelines for both code and models, through deployment and monitoring.
• Working With Juniors: Comfortable guiding junior engineers on a shared track, and delivering your own work to production as part of a team not building in isolation.
• Team Player (must-have): A genuine collaborator: communicative, reliable, and generous with help. You own delivery without working in a silo and raise blockers early.
Nice to Have
• Advanced ML: VLM fine-tuning (LoRA/QLoRA) or on-device / edge ML experience.
• Education / Recognition: Master’s or Ph.D. in a relevant field, publications (CVPR/ICCV/NeurIPS), or granted patents.
• Domain: Biomechanics, fitness/sports analytics, health-tech, or AR/VR.
What You’ll Gain
• Ownership: Own real models end to end training through real-time edge deployment on a cutting-edge human-movement platform.
• Deep-Tech Craft: Work on hard, latency-critical, privacy-first ML rather than dashboards and glue.
• Patent Creation: Direct involvement in building patentable core technology.
• Growth: A clear path to senior/lead ML as the platform and team scale.
How to Apply
Interested candidates can send a CV and a short note about relevant work ideally with links to code, papers, or shipped ML systems to careers@nutpaa.com.
Please use the subject line “Machine Learning Engineer.”
Location: Madurai, Tamil Nadu · Hybrid · Full-time.
Apply now to join us
About Nutpaa
Nutpaa is an early-stage deep-tech company building real-time human movement perception and intelligence using computer vision, machine learning, and 3D modeling. Our systems run on-device, combine multi-camera and wearable capture, and are built around low-latency inference and strict privacy protection.
Our culture centers on technical depth, strong engineering ownership, peer collaboration, long-term thinking, and patent-protected innovation. This is a role for an engineer who writes excellent code and takes models all the way from research to real-time production on the edge.
Role Summary
We are looking for a Machine Learning Engineer who is, first and foremost, a strong coder someone who builds robust, efficient, production-grade ML systems, not just notebooks. You will own models across our perception and coaching stack: training and evaluation, data and feature pipelines, and the optimization and deployment that get those models running in real time on edge hardware.
Machine learning is the heavy lifting in this role but our perception stack is camera-based, so you will also do solid, mid-level computer-vision engineering: image and video processing, camera calibration, multi-view geometry, and 2D/3D pose as the inputs your models depend on. Think ML-first, with CV as a strong working competency rather than a part-time interest.
You will work hands-on alongside our computer-vision, full-stack, and (as the team grows) edge and hardware engineers. The bar for code quality is high: clean, tested, reproducible, and fast. You will own your ML track and its quality, work alongside and guide junior ML engineers on it, and follow strong SDLC and CI/CD discipline across the ML lifecycle. We are looking for someone who ships tested work and collaborates well, not a solo developer.
Key Responsibilities
1. Model Development & Training
• Build & Train: Design, train, fine-tune, and evaluate models across pose/movement understanding, sequence/temporal modeling, and multimodal or Vision-Language tasks, using PyTorch.
• Rigorous Evaluation: Define metrics, benchmarks, and error analysis; drive measurable accuracy, robustness, and reliability improvements rather than one-off results.
• Data & Features: Build data and feature pipelines dataset preparation, augmentation, synthetic data, and labeling workflows that are versioned and reproducible.
• Applied Computer Vision: Build and integrate the camera-based perception around your models image/video processing (OpenCV), camera calibration, multi-view geometry, and 2D/3D pose at a solid mid-level. ML remains the core craft; CV is the working layer that feeds it.
2. Optimization & Edge Deployment
• Model Optimization: Quantize (INT8/FP16), prune, and distill models, and manage export to ONNX, TensorRT, TFLite, or CoreML for real-time inference.
• Edge Performance: Hit strict latency, memory, and thermal budgets on Jetson-class edge and NPU targets; profile and optimize the hot paths that matter.
• Production Inference: Ship models into a real serving path reproducible builds, versioning, monitoring, and clean rollbacks.
3. Engineering Excellence (Strong Coding)
• Production-Grade Code: Write clean, efficient, well-tested Python (and performance-sensitive C++ where needed); strong grasp of data structures, algorithms, and complexity.
• Software Discipline: Unit/integration tests, code reviews, profiling, clear documentation, Git, and CI/CD for models and pipelines.
• Reproducibility: Experiment tracking, deterministic pipelines, and MLOps practices so results are reproducible and deployable, not just demoable.
4. Track Ownership, Mentoring & Teamwork
• Own Your ML Track: Take ownership of your ML track the models and components assigned to you and deliver them well: designed, tested, and cleanly integrated, working closely with the wider team.
• Work With & Guide Juniors: Work alongside junior ML engineers on your track guide their design and code reviews, help grow their skills, and unblock them.
• Test Your Work: Own the quality of what you ship. Unit and integration tests, model validation, and regression checks are part of “done” not an afterthought. We expect engineers who test their work, not just develop it.
• SDLC on the ML Side: Own the software-development lifecycle for your ML work Git branching and pull requests, code reviews, automated testing, and CI/CD for both code and models, through deployment and monitoring.
• Team Player: Collaborate openly and reliably. Partner day-to-day with CV, full-stack, and edge/hardware engineers, communicate proactively, and flag dependencies and blockers early. This is a team sport, not a solo one.
What We Are Looking For
Experience & Coding
• Background: 4–6 years building and shipping machine-learning systems in production, with a solid record of taking models from research to deployment.
• Strong Coder (must-have): Excellent, demonstrable coding ability clean, efficient, production code; solid data structures, algorithms, and complexity fundamentals. A public code profile (GitHub / competitive programming) is a strong plus.
Machine Learning
• Core ML/DL: Deep, hands-on experience with PyTorch (and/or TensorFlow); CNNs, sequence/temporal models, and the training/evaluation lifecycle end to end.
• Applied Depth: Strength in one or more of: computer vision / pose estimation, multimodal or Vision-Language models, or real-time / time-series modeling.
Computer Vision (mid-level - required)
• Working CV: Solid, hands-on proficiency with OpenCV and image/video processing.
• Applied Geometry / Pose: Practical experience with camera calibration, multi-view geometry, or 2D/3D pose estimation. Deep specialization is not required but real, hands-on CV exposure is, since it feeds the models you build.
Optimization & Deployment
• Edge & Optimization: Hands-on with quantization and model optimization, and export/runtime tools (TensorRT, ONNX Runtime, TFLite, or CoreML).
• Targets: Experience deploying real-time models to GPUs and, ideally, edge/NPU or mobile hardware.
Foundations
• Toolkit: Python (NumPy, Pandas), strong debugging and profiling, Git, Linux, and Docker.
• MLOps: Experiment tracking, model versioning, and reproducible training/serving pipelines.
Engineering Discipline & Ways of Working
• Testing & Quality (must-have): A demonstrated habit of testing your own work unit and integration tests, model validation, and CI-gated quality checks. You treat untested code as unfinished.
• SDLC & CI/CD: Solid command of the software development lifecycle: Git branching and pull requests, code reviews, and CI/CD pipelines for both code and models, through deployment and monitoring.
• Working With Juniors: Comfortable guiding junior engineers on a shared track, and delivering your own work to production as part of a team not building in isolation.
• Team Player (must-have): A genuine collaborator: communicative, reliable, and generous with help. You own delivery without working in a silo and raise blockers early.
Nice to Have
• Advanced ML: VLM fine-tuning (LoRA/QLoRA) or on-device / edge ML experience.
• Education / Recognition: Master’s or Ph.D. in a relevant field, publications (CVPR/ICCV/NeurIPS), or granted patents.
• Domain: Biomechanics, fitness/sports analytics, health-tech, or AR/VR.
What You’ll Gain
• Ownership: Own real models end to end training through real-time edge deployment on a cutting-edge human-movement platform.
• Deep-Tech Craft: Work on hard, latency-critical, privacy-first ML rather than dashboards and glue.
• Patent Creation: Direct involvement in building patentable core technology.
• Growth: A clear path to senior/lead ML as the platform and team scale.
How to Apply
Interested candidates can send a CV and a short note about relevant work ideally with links to code, papers, or shipped ML systems to careers@nutpaa.com.
Please use the subject line “Machine Learning Engineer.”
Location: Madurai, Tamil Nadu · Hybrid · Full-time.
Apply now to join us
About Nutpaa
Nutpaa is an early-stage deep-tech company building real-time human movement perception and intelligence using computer vision, machine learning, and 3D modeling. Our systems run on-device, combine multi-camera and wearable capture, and are built around low-latency inference and strict privacy protection.
Our culture centers on technical depth, strong engineering ownership, peer collaboration, long-term thinking, and patent-protected innovation. This is a role for an engineer who writes excellent code and takes models all the way from research to real-time production on the edge.
Role Summary
We are looking for a Machine Learning Engineer who is, first and foremost, a strong coder someone who builds robust, efficient, production-grade ML systems, not just notebooks. You will own models across our perception and coaching stack: training and evaluation, data and feature pipelines, and the optimization and deployment that get those models running in real time on edge hardware.
Machine learning is the heavy lifting in this role but our perception stack is camera-based, so you will also do solid, mid-level computer-vision engineering: image and video processing, camera calibration, multi-view geometry, and 2D/3D pose as the inputs your models depend on. Think ML-first, with CV as a strong working competency rather than a part-time interest.
You will work hands-on alongside our computer-vision, full-stack, and (as the team grows) edge and hardware engineers. The bar for code quality is high: clean, tested, reproducible, and fast. You will own your ML track and its quality, work alongside and guide junior ML engineers on it, and follow strong SDLC and CI/CD discipline across the ML lifecycle. We are looking for someone who ships tested work and collaborates well, not a solo developer.
Key Responsibilities
1. Model Development & Training
• Build & Train: Design, train, fine-tune, and evaluate models across pose/movement understanding, sequence/temporal modeling, and multimodal or Vision-Language tasks, using PyTorch.
• Rigorous Evaluation: Define metrics, benchmarks, and error analysis; drive measurable accuracy, robustness, and reliability improvements rather than one-off results.
• Data & Features: Build data and feature pipelines dataset preparation, augmentation, synthetic data, and labeling workflows that are versioned and reproducible.
• Applied Computer Vision: Build and integrate the camera-based perception around your models image/video processing (OpenCV), camera calibration, multi-view geometry, and 2D/3D pose at a solid mid-level. ML remains the core craft; CV is the working layer that feeds it.
2. Optimization & Edge Deployment
• Model Optimization: Quantize (INT8/FP16), prune, and distill models, and manage export to ONNX, TensorRT, TFLite, or CoreML for real-time inference.
• Edge Performance: Hit strict latency, memory, and thermal budgets on Jetson-class edge and NPU targets; profile and optimize the hot paths that matter.
• Production Inference: Ship models into a real serving path reproducible builds, versioning, monitoring, and clean rollbacks.
3. Engineering Excellence (Strong Coding)
• Production-Grade Code: Write clean, efficient, well-tested Python (and performance-sensitive C++ where needed); strong grasp of data structures, algorithms, and complexity.
• Software Discipline: Unit/integration tests, code reviews, profiling, clear documentation, Git, and CI/CD for models and pipelines.
• Reproducibility: Experiment tracking, deterministic pipelines, and MLOps practices so results are reproducible and deployable, not just demoable.
4. Track Ownership, Mentoring & Teamwork
• Own Your ML Track: Take ownership of your ML track the models and components assigned to you and deliver them well: designed, tested, and cleanly integrated, working closely with the wider team.
• Work With & Guide Juniors: Work alongside junior ML engineers on your track guide their design and code reviews, help grow their skills, and unblock them.
• Test Your Work: Own the quality of what you ship. Unit and integration tests, model validation, and regression checks are part of “done” not an afterthought. We expect engineers who test their work, not just develop it.
• SDLC on the ML Side: Own the software-development lifecycle for your ML work Git branching and pull requests, code reviews, automated testing, and CI/CD for both code and models, through deployment and monitoring.
• Team Player: Collaborate openly and reliably. Partner day-to-day with CV, full-stack, and edge/hardware engineers, communicate proactively, and flag dependencies and blockers early. This is a team sport, not a solo one.
What We Are Looking For
Experience & Coding
• Background: 4–6 years building and shipping machine-learning systems in production, with a solid record of taking models from research to deployment.
• Strong Coder (must-have): Excellent, demonstrable coding ability clean, efficient, production code; solid data structures, algorithms, and complexity fundamentals. A public code profile (GitHub / competitive programming) is a strong plus.
Machine Learning
• Core ML/DL: Deep, hands-on experience with PyTorch (and/or TensorFlow); CNNs, sequence/temporal models, and the training/evaluation lifecycle end to end.
• Applied Depth: Strength in one or more of: computer vision / pose estimation, multimodal or Vision-Language models, or real-time / time-series modeling.
Computer Vision (mid-level - required)
• Working CV: Solid, hands-on proficiency with OpenCV and image/video processing.
• Applied Geometry / Pose: Practical experience with camera calibration, multi-view geometry, or 2D/3D pose estimation. Deep specialization is not required but real, hands-on CV exposure is, since it feeds the models you build.
Optimization & Deployment
• Edge & Optimization: Hands-on with quantization and model optimization, and export/runtime tools (TensorRT, ONNX Runtime, TFLite, or CoreML).
• Targets: Experience deploying real-time models to GPUs and, ideally, edge/NPU or mobile hardware.
Foundations
• Toolkit: Python (NumPy, Pandas), strong debugging and profiling, Git, Linux, and Docker.
• MLOps: Experiment tracking, model versioning, and reproducible training/serving pipelines.
Engineering Discipline & Ways of Working
• Testing & Quality (must-have): A demonstrated habit of testing your own work unit and integration tests, model validation, and CI-gated quality checks. You treat untested code as unfinished.
• SDLC & CI/CD: Solid command of the software development lifecycle: Git branching and pull requests, code reviews, and CI/CD pipelines for both code and models, through deployment and monitoring.
• Working With Juniors: Comfortable guiding junior engineers on a shared track, and delivering your own work to production as part of a team not building in isolation.
• Team Player (must-have): A genuine collaborator: communicative, reliable, and generous with help. You own delivery without working in a silo and raise blockers early.
Nice to Have
• Advanced ML: VLM fine-tuning (LoRA/QLoRA) or on-device / edge ML experience.
• Education / Recognition: Master’s or Ph.D. in a relevant field, publications (CVPR/ICCV/NeurIPS), or granted patents.
• Domain: Biomechanics, fitness/sports analytics, health-tech, or AR/VR.
What You’ll Gain
• Ownership: Own real models end to end training through real-time edge deployment on a cutting-edge human-movement platform.
• Deep-Tech Craft: Work on hard, latency-critical, privacy-first ML rather than dashboards and glue.
• Patent Creation: Direct involvement in building patentable core technology.
• Growth: A clear path to senior/lead ML as the platform and team scale.
How to Apply
Interested candidates can send a CV and a short note about relevant work ideally with links to code, papers, or shipped ML systems to careers@nutpaa.com.
Please use the subject line “Machine Learning Engineer.”
Location: Madurai, Tamil Nadu · Hybrid · Full-time.
About Nutpaa
Nutpaa is an early-stage deep-tech company building real-time human movement perception and intelligence using computer vision, machine learning, and 3D modeling. Our systems run on-device, combine multi-camera and wearable capture, and are built around low-latency inference and strict privacy protection.
Our culture centers on technical depth, strong engineering ownership, peer collaboration, long-term thinking, and patent-protected innovation. This is a role for an engineer who writes excellent code and takes models all the way from research to real-time production on the edge.
Role Summary
We are looking for a Machine Learning Engineer who is, first and foremost, a strong coder someone who builds robust, efficient, production-grade ML systems, not just notebooks. You will own models across our perception and coaching stack: training and evaluation, data and feature pipelines, and the optimization and deployment that get those models running in real time on edge hardware.
Machine learning is the heavy lifting in this role but our perception stack is camera-based, so you will also do solid, mid-level computer-vision engineering: image and video processing, camera calibration, multi-view geometry, and 2D/3D pose as the inputs your models depend on. Think ML-first, with CV as a strong working competency rather than a part-time interest.
You will work hands-on alongside our computer-vision, full-stack, and (as the team grows) edge and hardware engineers. The bar for code quality is high: clean, tested, reproducible, and fast. You will own your ML track and its quality, work alongside and guide junior ML engineers on it, and follow strong SDLC and CI/CD discipline across the ML lifecycle. We are looking for someone who ships tested work and collaborates well, not a solo developer.
Key Responsibilities
1. Model Development & Training
• Build & Train: Design, train, fine-tune, and evaluate models across pose/movement understanding, sequence/temporal modeling, and multimodal or Vision-Language tasks, using PyTorch.
• Rigorous Evaluation: Define metrics, benchmarks, and error analysis; drive measurable accuracy, robustness, and reliability improvements rather than one-off results.
• Data & Features: Build data and feature pipelines dataset preparation, augmentation, synthetic data, and labeling workflows that are versioned and reproducible.
• Applied Computer Vision: Build and integrate the camera-based perception around your models image/video processing (OpenCV), camera calibration, multi-view geometry, and 2D/3D pose at a solid mid-level. ML remains the core craft; CV is the working layer that feeds it.
2. Optimization & Edge Deployment
• Model Optimization: Quantize (INT8/FP16), prune, and distill models, and manage export to ONNX, TensorRT, TFLite, or CoreML for real-time inference.
• Edge Performance: Hit strict latency, memory, and thermal budgets on Jetson-class edge and NPU targets; profile and optimize the hot paths that matter.
• Production Inference: Ship models into a real serving path reproducible builds, versioning, monitoring, and clean rollbacks.
3. Engineering Excellence (Strong Coding)
• Production-Grade Code: Write clean, efficient, well-tested Python (and performance-sensitive C++ where needed); strong grasp of data structures, algorithms, and complexity.
• Software Discipline: Unit/integration tests, code reviews, profiling, clear documentation, Git, and CI/CD for models and pipelines.
• Reproducibility: Experiment tracking, deterministic pipelines, and MLOps practices so results are reproducible and deployable, not just demoable.
4. Track Ownership, Mentoring & Teamwork
• Own Your ML Track: Take ownership of your ML track the models and components assigned to you and deliver them well: designed, tested, and cleanly integrated, working closely with the wider team.
• Work With & Guide Juniors: Work alongside junior ML engineers on your track guide their design and code reviews, help grow their skills, and unblock them.
• Test Your Work: Own the quality of what you ship. Unit and integration tests, model validation, and regression checks are part of “done” not an afterthought. We expect engineers who test their work, not just develop it.
• SDLC on the ML Side: Own the software-development lifecycle for your ML work Git branching and pull requests, code reviews, automated testing, and CI/CD for both code and models, through deployment and monitoring.
• Team Player: Collaborate openly and reliably. Partner day-to-day with CV, full-stack, and edge/hardware engineers, communicate proactively, and flag dependencies and blockers early. This is a team sport, not a solo one.
What We Are Looking For
Experience & Coding
• Background: 4–6 years building and shipping machine-learning systems in production, with a solid record of taking models from research to deployment.
• Strong Coder (must-have): Excellent, demonstrable coding ability clean, efficient, production code; solid data structures, algorithms, and complexity fundamentals. A public code profile (GitHub / competitive programming) is a strong plus.
Machine Learning
• Core ML/DL: Deep, hands-on experience with PyTorch (and/or TensorFlow); CNNs, sequence/temporal models, and the training/evaluation lifecycle end to end.
• Applied Depth: Strength in one or more of: computer vision / pose estimation, multimodal or Vision-Language models, or real-time / time-series modeling.
Computer Vision (mid-level - required)
• Working CV: Solid, hands-on proficiency with OpenCV and image/video processing.
• Applied Geometry / Pose: Practical experience with camera calibration, multi-view geometry, or 2D/3D pose estimation. Deep specialization is not required but real, hands-on CV exposure is, since it feeds the models you build.
Optimization & Deployment
• Edge & Optimization: Hands-on with quantization and model optimization, and export/runtime tools (TensorRT, ONNX Runtime, TFLite, or CoreML).
• Targets: Experience deploying real-time models to GPUs and, ideally, edge/NPU or mobile hardware.
Foundations
• Toolkit: Python (NumPy, Pandas), strong debugging and profiling, Git, Linux, and Docker.
• MLOps: Experiment tracking, model versioning, and reproducible training/serving pipelines.
Engineering Discipline & Ways of Working
• Testing & Quality (must-have): A demonstrated habit of testing your own work unit and integration tests, model validation, and CI-gated quality checks. You treat untested code as unfinished.
• SDLC & CI/CD: Solid command of the software development lifecycle: Git branching and pull requests, code reviews, and CI/CD pipelines for both code and models, through deployment and monitoring.
• Working With Juniors: Comfortable guiding junior engineers on a shared track, and delivering your own work to production as part of a team not building in isolation.
• Team Player (must-have): A genuine collaborator: communicative, reliable, and generous with help. You own delivery without working in a silo and raise blockers early.
Nice to Have
• Advanced ML: VLM fine-tuning (LoRA/QLoRA) or on-device / edge ML experience.
• Education / Recognition: Master’s or Ph.D. in a relevant field, publications (CVPR/ICCV/NeurIPS), or granted patents.
• Domain: Biomechanics, fitness/sports analytics, health-tech, or AR/VR.
What You’ll Gain
• Ownership: Own real models end to end training through real-time edge deployment on a cutting-edge human-movement platform.
• Deep-Tech Craft: Work on hard, latency-critical, privacy-first ML rather than dashboards and glue.
• Patent Creation: Direct involvement in building patentable core technology.
• Growth: A clear path to senior/lead ML as the platform and team scale.
How to Apply
Interested candidates can send a CV and a short note about relevant work ideally with links to code, papers, or shipped ML systems to careers@nutpaa.com.
Please use the subject line “Machine Learning Engineer.”
Location: Madurai, Tamil Nadu · Hybrid · Full-time.
Apply now to join us
Apply now to join us
Madurai, Tamil Nadu · Hybrid
4–6 years of experience
About Nutpaa
Nutpaa is an early-stage deep-tech company building real-time human movement perception and intelligence using computer vision, machine learning, and 3D modeling. Our systems run on-device, combine multi-camera and wearable capture, and are built around low-latency inference and strict privacy protection.
Our culture centers on technical depth, strong engineering ownership, peer collaboration, long-term thinking, and patent-protected innovation. This is a role for an engineer who writes excellent code and takes models all the way from research to real-time production on the edge.
Role Summary
We are looking for a Machine Learning Engineer who is, first and foremost, a strong coder someone who builds robust, efficient, production-grade ML systems, not just notebooks. You will own models across our perception and coaching stack: training and evaluation, data and feature pipelines, and the optimization and deployment that get those models running in real time on edge hardware.
Machine learning is the heavy lifting in this role but our perception stack is camera-based, so you will also do solid, mid-level computer-vision engineering: image and video processing, camera calibration, multi-view geometry, and 2D/3D pose as the inputs your models depend on. Think ML-first, with CV as a strong working competency rather than a part-time interest.
You will work hands-on alongside our computer-vision, full-stack, and (as the team grows) edge and hardware engineers. The bar for code quality is high: clean, tested, reproducible, and fast. You will own your ML track and its quality, work alongside and guide junior ML engineers on it, and follow strong SDLC and CI/CD discipline across the ML lifecycle. We are looking for someone who ships tested work and collaborates well, not a solo developer.
Key Responsibilities
1. Model Development & Training
• Build & Train: Design, train, fine-tune, and evaluate models across pose/movement understanding, sequence/temporal modeling, and multimodal or Vision-Language tasks, using PyTorch.
• Rigorous Evaluation: Define metrics, benchmarks, and error analysis; drive measurable accuracy, robustness, and reliability improvements rather than one-off results.
• Data & Features: Build data and feature pipelines dataset preparation, augmentation, synthetic data, and labeling workflows that are versioned and reproducible.
• Applied Computer Vision: Build and integrate the camera-based perception around your models image/video processing (OpenCV), camera calibration, multi-view geometry, and 2D/3D pose at a solid mid-level. ML remains the core craft; CV is the working layer that feeds it.
2. Optimization & Edge Deployment
• Model Optimization: Quantize (INT8/FP16), prune, and distill models, and manage export to ONNX, TensorRT, TFLite, or CoreML for real-time inference.
• Edge Performance: Hit strict latency, memory, and thermal budgets on Jetson-class edge and NPU targets; profile and optimize the hot paths that matter.
• Production Inference: Ship models into a real serving path reproducible builds, versioning, monitoring, and clean rollbacks.
3. Engineering Excellence (Strong Coding)
• Production-Grade Code: Write clean, efficient, well-tested Python (and performance-sensitive C++ where needed); strong grasp of data structures, algorithms, and complexity.
• Software Discipline: Unit/integration tests, code reviews, profiling, clear documentation, Git, and CI/CD for models and pipelines.
• Reproducibility: Experiment tracking, deterministic pipelines, and MLOps practices so results are reproducible and deployable, not just demoable.
4. Track Ownership, Mentoring & Teamwork
• Own Your ML Track: Take ownership of your ML track the models and components assigned to you and deliver them well: designed, tested, and cleanly integrated, working closely with the wider team.
• Work With & Guide Juniors: Work alongside junior ML engineers on your track guide their design and code reviews, help grow their skills, and unblock them.
• Test Your Work: Own the quality of what you ship. Unit and integration tests, model validation, and regression checks are part of “done” not an afterthought. We expect engineers who test their work, not just develop it.
• SDLC on the ML Side: Own the software-development lifecycle for your ML work Git branching and pull requests, code reviews, automated testing, and CI/CD for both code and models, through deployment and monitoring.
• Team Player: Collaborate openly and reliably. Partner day-to-day with CV, full-stack, and edge/hardware engineers, communicate proactively, and flag dependencies and blockers early. This is a team sport, not a solo one.
What We Are Looking For
Experience & Coding
• Background: 4–6 years building and shipping machine-learning systems in production, with a solid record of taking models from research to deployment.
• Strong Coder (must-have): Excellent, demonstrable coding ability clean, efficient, production code; solid data structures, algorithms, and complexity fundamentals. A public code profile (GitHub / competitive programming) is a strong plus.
Machine Learning
• Core ML/DL: Deep, hands-on experience with PyTorch (and/or TensorFlow); CNNs, sequence/temporal models, and the training/evaluation lifecycle end to end.
• Applied Depth: Strength in one or more of: computer vision / pose estimation, multimodal or Vision-Language models, or real-time / time-series modeling.
Computer Vision (mid-level - required)
• Working CV: Solid, hands-on proficiency with OpenCV and image/video processing.
• Applied Geometry / Pose: Practical experience with camera calibration, multi-view geometry, or 2D/3D pose estimation. Deep specialization is not required but real, hands-on CV exposure is, since it feeds the models you build.
Optimization & Deployment
• Edge & Optimization: Hands-on with quantization and model optimization, and export/runtime tools (TensorRT, ONNX Runtime, TFLite, or CoreML).
• Targets: Experience deploying real-time models to GPUs and, ideally, edge/NPU or mobile hardware.
Foundations
• Toolkit: Python (NumPy, Pandas), strong debugging and profiling, Git, Linux, and Docker.
• MLOps: Experiment tracking, model versioning, and reproducible training/serving pipelines.
Engineering Discipline & Ways of Working
• Testing & Quality (must-have): A demonstrated habit of testing your own work unit and integration tests, model validation, and CI-gated quality checks. You treat untested code as unfinished.
• SDLC & CI/CD: Solid command of the software development lifecycle: Git branching and pull requests, code reviews, and CI/CD pipelines for both code and models, through deployment and monitoring.
• Working With Juniors: Comfortable guiding junior engineers on a shared track, and delivering your own work to production as part of a team not building in isolation.
• Team Player (must-have): A genuine collaborator: communicative, reliable, and generous with help. You own delivery without working in a silo and raise blockers early.
Nice to Have
• Advanced ML: VLM fine-tuning (LoRA/QLoRA) or on-device / edge ML experience.
• Education / Recognition: Master’s or Ph.D. in a relevant field, publications (CVPR/ICCV/NeurIPS), or granted patents.
• Domain: Biomechanics, fitness/sports analytics, health-tech, or AR/VR.
What You’ll Gain
• Ownership: Own real models end to end training through real-time edge deployment on a cutting-edge human-movement platform.
• Deep-Tech Craft: Work on hard, latency-critical, privacy-first ML rather than dashboards and glue.
• Patent Creation: Direct involvement in building patentable core technology.
• Growth: A clear path to senior/lead ML as the platform and team scale.
How to Apply
Interested candidates can send a CV and a short note about relevant work ideally with links to code, papers, or shipped ML systems to careers@nutpaa.com.
Please use the subject line “Machine Learning Engineer.”
Location: Madurai, Tamil Nadu · Hybrid · Full-time.
Apply now to join us