5 hours ago
Zürich, SwitzerlandIntern
Responsibilities
- Train reinforcement-learning policies for navigation, obstacle avoidance, dynamic goal-seeking, and autonomous flight in complex 3D environments.
- Combine learned cost shaping and value functions with trajectory optimization for safe, smooth, agile flight.
- Build scalable simulation datasets and training loops using domain randomization, residual learning, and safety filters, then validate on real drones.
- Develop human-in-the-loop shared-control policies that blend pilot intent, autonomy priors, and uncertainty-aware behavior.
- Explore decentralized multi-agent coordination for coverage, pursuit, and collaborative mapping.
- Prototype in simulation and flight-test policies on production-ready aircraft and edge compute.
- Collaborate with perception, mapping, controls, and product teams to transition research into field operations.
Requirements
- Be a PhD student in Robotics, Machine Learning, Controls, or a related field.
- Have strong fundamentals in reinforcement learning, control theory, and motion planning, with familiarity with safety and robustness concepts.
- Be proficient in Python and PyTorch, JAX, or Ray RLlib, and in at least one of C++ or CUDA.
- Have hands-on experience with robotics simulation using Isaac Lab, MuJoCo, or PyBullet and with sim-to-real techniques.
- Have experience training or deploying policies for navigation, manipulation, or locomotion on real robots or autonomous vehicles.
- Publications at venues such as CoRL, ICRA, IROS, RSS, or NeurIPS are a nice-to-have.
- Experience with onboard inference optimization, TensorRT, quantization, sparsity, diffusion policies, imitation learning or behavior cloning, offline RL, model-based RL, multi-agent RL, or distributed training is preferred.
Benefits
- Hourly compensation is $47/hr for undergraduate students, $53/hr for master's students, and $70/hr for PhD students.
- Work includes frequent validation on real drones and production-ready aircraft, with mentorship across perception, mapping, controls, and product teams.
- Skydio full-time employees are eligible to enroll in benefit plans and access various incentives and stipends.
Categories
ML EngineeringRobotics
About Skydio
Skydio builds autonomous drones, docking systems, and fleet management software for enterprise, public safety, and defense customers. It sells hardware with AI-based autonomy plus SaaS for fleet operations, inspections, and Drone-as-First-Responder programs. Founded in 2014 and headquartered in San Mateo, California, the company designs and assembles its products in the U.S., with manufacturing in Hayward, and its systems are used across utilities, law enforcement, and the U.S. military.