Robotics Software Engineer, Mission Evaluation & Fine-Tuning

FieldAI • Irvine, CA

4d • Onsite

About The Position

FieldAI’s Irvine team develops risk-aware, reliable robotics AI for real-world deployments. This role owns the end-to-end inspection behavior success metric and builds evaluation processes that use labeled datasets, seeded-fault runs, and regression suites to gate release candidates. The engineer turns field evidence into training and evaluation sets, connects technical performance to customer-measured metrics, and makes failures understandable by analyzing mission logs, identifying their source, and routing evidence to the responsible team.

The position involves evaluating and fine-tuning vision or vision-language models, developing the datasets and metrics that support them, and building data pipelines that move evidence from robots into training sets and dashboards. Work is performed in collaboration with AI, software, robotics engineering, product, field deployment, and technical communication teams. The Irvine team tests systems on real hardware and sensors and ships them for operation in unstructured environments.

Requirements

• Master’s degree in computer science, robotics, aerospace engineering, or a related field.
• At least 3 years of experience in robotics or ML engineering.
• Strong Python skills.
• Working knowledge of C++.
• Comfort with ROS 2 logs, bags, and replay tooling.
• Experience evaluating and fine-tuning vision or vision-language models, using tools such as PyTorch and LoRA or similar methods, including building datasets and metrics.
• Statistical rigor, including understanding the distinction between benchmark numbers and field results and designing experiments to distinguish them.
• Data pipeline skills to move evidence from robot to training set to dashboard.

Nice To Haves

• Authored or maintained a robotics benchmark, evaluation suite, or challenge.
• Simulation-based evaluation and sim-to-real validation experience.
• MLOps experience for on-edge models, including quantization, versioning, and fleet rollout.
• Experience in industrial inspection, gauge reading, or anomaly detection.

Responsibilities

• Own the end-to-end inspection behavior success metric.
• Build evaluation loops using golden-labeled datasets, seeded-fault runs, and regression suites at feature, system, and release levels to gate release candidates.
• Turn field evidence into training and evaluation sets.
• Map technical performance to customer-measured metrics.
• Triage mission logs, attribute misses to navigation, localization, perception, or behavior, and route them to the owning team with supporting evidence.
• Evaluate and fine-tune vision or vision-language models and build the datasets and metrics behind them.
• Build data pipelines that move evidence from robots to training sets and dashboards.

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level