What are the responsibilities and job description for the ML Training Engineer position at TerraFirma Robotics?
About TerraFirma Robotics
TerraFirma Robotics is developing autonomous drone-based systems and data products for Earth monitoring and environmental intelligence. Our work spans robotics, edge AI, sensing, autonomy, and digital twin infrastructure. We care about technical rigor, thoughtful execution, and building systems that are useful in the real world, not just impressive in a demo.
About the Role
This role is for an engineer who wants to contribute meaningfully in a small, highly collaborative environment. You’ll work on core technical problems, help shape implementation decisions, and partner across disciplines to move systems from concept to reliable execution. We value clear thinking, sound engineering judgment, and a willingness to work across boundaries when needed.
Compensation: Full-time salary equity option
Start Date: May 15, 2026
The Role
You train the models that the AI/Perception Engineer architects. You build datasets, run training pipelines, manage experiments, and deliver INT8-quantized models ready for Jetson deployment. You help turn sparse early data and synthetic pipelines into deployable perception models that improve with each iteration.
What You'll Do
- Train safety detection models using aerial imagery datasets plus synthetic data augmentation
- Fine-tune multimodal and vision-language models for environmental scene understanding
- Develop edge-deployable model variants for constrained compute environments
- Implement confidence calibration for safety-critical predictions
- Manage experiment tracking, dataset versioning, and model validation pipelines
- Build export-ready training workflows for ONNX and TensorRT deployment
Requirements
- PyTorch model training at scale (multi-GPU, distributed)
- NVIDIA TAO Toolkit or equivalent transfer-learning workflow
- Dataset management: annotation pipelines, versioning, class balancing, and QA
- Model quantization: PTQ/QAT for INT8/FP16, TensorRT export
- Experiment tracking (MLflow, Weights & Biases, or similar)
- Azure GPU infrastructure management
- Understanding of precision/recall tradeoffs and calibration for safety-critical systems
Preferred
- Aerial/drone imagery training (VisDrone, DOTA datasets)
- NVIDIA Isaac Sim for synthetic data generation
- VQ-VAE or neural compression model training
- Confidence calibration techniques
- Embedded deployment targets (Jetson, edge TPU)
- Agricultural or environmental remote sensing domain knowledge
Why TerraFirma
- Meaningful technical ownership from the start
- Close collaboration with a small, high-trust team
- Opportunity to work on multidisciplinary problems across robotics, AI, sensing, and systems
- A chance to help shape products and technical foundations early
- Competitive compensation, with benefits structured by role and hiring location
Compensation and benefits vary by role, level, and location, and will be discussed during the process.
Equal Opportunity
TerraFirma Robotics is an equal opportunity employer. We evaluate candidates based on qualifications, capability, and potential to contribute. We do not discriminate on the basis of race, color, religion, sex, gender identity or expression, sexual orientation, national origin, age, disability, veteran status, or other protected status under applicable law.