Find The RightJob.

Staff ML Engineer, Perception Scene Understanding

Failsafe is a foundational pillar of the Autonomy safety architecture. In this Staff ML Software Engineer role, you will play a key part in driving and delivering high-quality, real-time onboard models and algorithms to evaluate sensor reliability, scene hazards, and perception capabilities. You will design, train, and ship production-grade solutions that enable our vehicles to understand environmental limitations, sensor degradations, and safety boundaries in 3D.

As an experienced technical expert, you will own the end-to-end lifecycle of the failsafe system: defining safety metrics, engineering diagnostic algorithms, benchmarking on-vehicle performance, and establishing tight feedback loops. A key focus of this role is dedicated to multi-modal scene understanding, ensuring the vehicle can gracefully handle degraded states (such as lens blockage, adverse weather, or road surface hazards) to trigger automated sensor recoverability routines or minimum risk maneuvers.

Lead the design, delivery and deployment of production-grade, highly reliable failsafe algorithms for L3 and L4 platforms, focusing on real-time sensor blockage, degradation, and occlusion detection.
Develop multi-modal (camera and/or LiDAR) models to analyze real-time road conditions, including surface wetness, friction/mu-estimation, and physical drivability hazards.
Design intelligent algorithms to trigger automated sensor-cleaning hardware and recovery strategies, directly maximizing ADAS uptime in challenging environments
Architect systems to evaluate and report real-time 3D perception uncertainty and capability limits, interfacing directly with downstream planning and central safety fallback managers.
Partner closely with cross-functional teams to define strict safety boundaries, establish rigorous testing benchmarks, and guide system-level architecture trade-offs.

BS, MS, or PhD in Computer Science, Robotics, Electrical Engineering, or a highly related quantitative field.
7+ years of professional experience scaling production-grade MLsoftware for autonomy or robotics.
Proven track record of hands-on experience building algorithms for autonomous vehicle perception or environmental scene understanding.
Solid understanding of modern perception pipelines and data representation across multiple modalities (cameras, LiDAR, and Radar).
Strong proficiency in C++ and Python alongside experience with real-time software performance optimization and latency profiling.
Demonstrated ability to drive complex technical roadmaps and deliver components in a fast-paced environment.

Preferred Qualifications

Experience designing failsafe, fail-operable, or degraded-state software architectures for safety-critical systems.
Hands-on experience with sensor degradation modeling (such as camera soil, lens glare, rain, or LiDAR spray) and system-level ODD degradation modeling.
Deep understanding of onboard edge deployment and managing cloud telemetry loops to monitor model performance in the field.
Experience collaborating directly with systems safety teams to translate hazard analysis and risk assessments into robust software implementations.

Similar jobs