Build pipelines that use deployment successes, failures, and teleoperation corrections for targeted fine-tuning and online policy improvement.
Analyze fleet sensor and video data to identify failure modes, drift, and regressions.
Apply reinforcement learning and feedback-based reward modeling to improve deployed robot policies.
Build real-time fleet monitoring for anomalies, near-failures, and out-of-distribution scenes, including decisions about when to involve a human.
Summarised from the posting by AI and can be wrong. The full description is the source.
Requirements
Experience
Hands-on experience in at least two of reinforcement learning, sensor-data modeling/anomaly detection, vision-language models, or continual/online learning
Experience building monitoring, evaluation, or data pipelines for a live ML system
Experience designing and reading production experiments such as A/B tests, canary rollouts, or staged fleet deployments, with statistical rigor to distinguish regressions from noise
Track record of hands-on effort shipped into a real system
Experience with robot fleets or other physically deployed autonomous systems in the field (preferred)
Experience building or fine-tuning perception or foundation models for automated monitoring, captioning, or anomaly detection (preferred)
Human-in-the-loop learning, including reward modeling from operator corrections, active learning, or failure-case data curation (preferred)
Education
Bachelor’s, Master’s, or PhD in CS, Robotics, Statistics, or a related field, or equivalent practical experience
Skills
Strong Python and PyTorch or JAX; comfort with large multimodal datasets and distributed compute such as Slurm/GPU clusters
Ability to turn a fleet-scale data investigation into a clear recommendation researchers and operators can act on
Statistical anomaly and drift detection methods, such as change-point detection or forecasting, applied to sensor or telemetry data (preferred)
Your match
Sign in and add your resume to see which of this job's requirements it shows.
Saved to your account, encrypted. The AI that reads it doesn't keep it or train on it.
Full job description
From Dyna Robotics’ careers site
Dyna Robotics builds general-purpose robots powered by a proprietary embodied AI foundation model with top-in-industry generalization and real-world performance. Already deployed with customers across multiple industries, our robots do commercial-grade work in the physical world. Our team comes from Google DeepMind, Meta, and Cruise, and we're backed by CRV, First Round, and other leading investors.
Show full descriptionShow less
As an Applied Researcher on the AI Research team, you’ll own the gap between “works in the lab” and “works at a new customer site, forever”: mining fleet sensor and video data for failure modes, building the monitoring that catches problems before customers do, and turning deployment data into continuous, measurable model improvement. Our models don’t stop learning at deployment. A growing fleet of robots is generating real production data every day, and closing that gap is a research problem, not just an ops one. This is a hands-on, ship-it role. We care far more about whether you can land a real improvement on the fleet than about producing research for its own sake.
This is a full-time, on-site position in Redwood City.
WHAT YOU’LL DO
Continuous Learning Loops: Design and ship pipelines that turn real deployment data (successes, failures, teleop corrections) into targeted fine-tuning and online policy improvement, closing the loop from field to model without a full retrain cycle every time.
Fleet Data Analytics: Mine high-frequency multimodal sensor and video data across tens of thousands of fleet episodes to catch failure modes, drift, and regressions before they become customer-visible.
RL for Deployment: Apply reinforcement learning (offline RL, RL fine-tuning, reward modeling from human and teleop feedback) to improve policies directly from real-world deployment data, not just simulation.
Automated Fleet Monitoring: Build automated monitoring that flags anomalies, near-failures, and out-of-distribution scenes across the fleet in real time, and that decides what needs a human versus what the system can self-correct.
Cross-Scene Generalization: Characterize and close generalization gaps as robots move to new sites, lighting, layouts, and objects; build the evaluation harnesses and data-selection strategies that make day-one performance at a new customer site predictable.
End-to-End Ownership: Partner with Research, Data, and Deployment teams to turn a finding into a shipped improvement, from a data-analysis notebook to a production monitoring dashboard to a deployed model update.
WHAT YOU’LL BRING
Bias to Ship: You’re happiest closing the loop, landing a fix on the fleet and watching it hold up at a real customer site, rather than polishing a benchmark number or a paper. We want someone driven by shipped impact and genuine passion for the problem, not research for its own sake.
Education: Bachelor’s, Master’s, or PhD in CS, Robotics, Statistics, or a related field, or equivalent practical experience. Degree level doesn’t matter to us; what matters is genuine passion for the work and a track record of hands-on effort that shipped into a real system, not just a benchmark.
Applied ML Depth: Hands-on experience in at least two of: reinforcement learning, sensor-data modeling/anomaly detection, vision-language models, or continual/online learning.
Production Instincts: Experience building monitoring, evaluation, or data pipelines for a live ML system, comfortable with the ambiguity of real-world fleet data versus curated benchmarks.
Experimentation & Statistics: Comfortable designing and reading production experiments (A/B tests, canary rollouts, staged fleet deployments) and applying enough statistical rigor to tell a real regression from noise in messy real-world data.
Technical Stack: Strong Python and PyTorch (or JAX); comfortable with large multimodal datasets and distributed compute (Slurm/GPU clusters).
Communication: Able to turn a fleet-scale data investigation into a clear recommendation that researchers and operators can act on.
BONUS POINTS FOR
Fleet Experience: Experience with robot fleets or other physically-deployed autonomous systems in the field, not just simulation.
Perception & Foundation Models: Experience building or fine-tuning perception or foundation models for automated monitoring, captioning, or anomaly detection.
Anomaly & Drift Detection: Background in statistical methods for detecting anomalies and drift (change-point detection, forecasting) applied to sensor or telemetry data.
Human-in-the-Loop Learning: Experience with human-in-the-loop learning: reward modeling from operator corrections, active learning, or data curation from failure cases.
Don’t let a checklist stop you. Data shows that underrepresented groups often only apply if they meet 100% of the criteria. We value problem-solving and grit over keyword matching. If you’re passionate about robotics, no matter your discipline, we want to hear from you, even if you don't check every box.
At Dyna Robotics, we build technology for the real world, which requires a team as diverse as the environments our robots inhabit. We are an equal opportunity employer committed to technical rigor and mutual respect.