Senior Machine Learning Engineer
Aerobotics7
IndiaremotePosted 7 days ago
Skill Required
Senior-ML-EngineerMachine-Learning-EngineerComputer-Vision-EngineerAI-EngineerSenior-AI-ML-EngineerSenior-ML-EngineeringMachine Learning EngineerMachine LearningSignal ProcessingETLObservabilitybuildingPyTorchwrittenPythonMLflowGitC++CloudComputer VisionAWSGCPandGenerative AIAIFulltime
Key highlights
- Compensation: ₹18-30 LPA CTC, based on skills and experience.
- Required experience: 5+ years applied ML shipping models into production.
- Key benefit: Strong stock options (equity) based on eligibility, performance and tenure.
- Notable requirement: Deep hands‑on experience with modern transformer‑based detection (DETR family, DINO‑DETR, etc.).
- Remote, India‑based full‑time role with daily US Pacific time overlap.
- Immediate start.
Role overview
The Senior Machine Learning Engineer will own the entire ML system at Aerobotics7, handling everything from limited real and synthetic datasets to real‑time models deployed on robots. As the sole ML engineer, you will set technical direction, build data pipelines, models, evaluation, deployment stacks, define the ML roadmap, and eventually hire and lead the growing ML team.
Responsibilities
- Own every ML system at A7 end-to-end - from limited real + synthetic datasets to models running in real time on the robot - as the only ML engineer, setting the technical direction for our next phase of development and deployment.
- You build the data pipeline, the models, the evaluation, and the deployment stack.
- You define the ML roadmap and, as we grow, hire and lead the team behind it.
- The full ML lifecycle: custom model development, data to training to evaluation to edge/cloud deployment to monitoring to iteration.
- Object detection and classification on hard, low-signal sensor data (radar, LiDAR, camera), including multi-sensor fusion.
- Our synthetic data strategy: generation, sim-to-real domain gap, domain adaptation and randomization.
- Real-time inference on-device (NVIDIA Jetson AGX Orin) and at scale in the cloud.
- ML infrastructure from scratch: experiment tracking, dataset/model versioning, reproducible training, evaluation harnesses.
- The ML roadmap. You set priorities with the founders and are accountable for outcomes, not experiments.
Requirements
- 5+ years applied ML shipping models into production, not research-only.
- Deep, hands-on work with modern transformer-based detection: DETR family (Deformable DETR, DINO-DETR), self-supervised backbones (DINOv2), or equivalent and newer architectures. You can explain the tradeoffs and when to reach for each. Off-the-shelf YOLO fine-tuning or LLM-generated pipelines you can't defend line by line are not what we're looking for.
- Low-data expertise: transfer learning, self-supervised / semi-supervised / few-shot, augmentation, active learning. You know how to get a strong model from a small labeled set.
- Synthetic-data training: you've trained on simulated data and closed the sim-to-real gap in a real system.
- Edge deployment: NVIDIA Jetson / AGX Orin, TensorRT, ONNX, quantization/pruning, latency and memory optimization.
- PyTorch fluency and strong Python.
- MLOps ownership: experiment tracking (W&B/MLflow), data/model versioning (e.g. DVC), reproducible pipelines.
- Self-direction: comfortable as the only ML person, working through ambiguity, owning decisions.
- Strong async written communication, and several hours of daily overlap with US Pacific time.
Nice to have
- Ground-penetrating radar, radar/signal processing, or geophysics.
- 3D / point-cloud ML.
- Cloud training and serving (AWS or GCP).
- C++ for edge/performance work.
- ROS2 or robotics/perception exposure.
- Track record of growing into a team lead.
Benefits
- Remote work (India‑based) full‑time.
- Strong stock options (equity) based on eligibility, performance and tenure.
- Opportunity to build and lead the ML organization from its first engineer.
- Work with unique, hard‑problem datasets and real robots in the field.
Additional details
- About the Company: Aerobotics7 (A7) is a growing, stealth‑stage startup building AI and robotics to solve some of the critical challenges on the planet. Small team, high ownership, real hardware in the field.
- Why this role is different: You are the ML function. What you build ships to a robot in the field, not a slide. You'll have datasets that don't exist anywhere else, hard problems worth solving, and the autonomy to solve them your way. If you want scope, ownership, and the chance to build an ML org from its first engineer up, this is that seat.
- Logistics: Location: Remote, India‑based. Hours: Daily overlap with US Pacific time required. Start: Immediate. Compensation: ₹18-30 LPA CTC, based on skills and experience. Equity: strong stock options based on eligibility, performance and tenure.
- Note: This role sits under Aerobotics7 Inventions Pvt. Ltd., our Indian entity. Compensation is aligned to Indian market standards. Our parent company is US‑based; this position is for candidates residing and working in India.
- How to Apply: Apply in Dover (Aerobotics7/41e08db2-c5a4-4bb6-be52-36c81ed012cd?rs=42706078) using your resume plus a GitHub or portfolio link (required). If any question email us at . A short note on relevant work is welcome but optional. If your best work is closed‑source, tell us what you built and what you owned.
- Originally posted on Himalayas.