Staff Perception Engineer, Pedestrian Detection

Senior Electrical Engineer

Location: San Diego, CA (On-site)

Type: Full-time

About Saferide

At Saferide, we are solving one of the hardest and most meaningful challenges in the world: ending road deaths. Our plug-and-play vision compute platform retrofits the 200M vehicles already on the road with advanced driver assistance and safety features, bringing the future of mobility to every driver, not just those buying new cars.

We are scaling from prototype to production, building the first real vision-based solution that can prevent the crashes that kill more than a million people globally each year. This is your chance to join a small, fast-moving team at the exact moment when our product goes from concept to reality.

The Role

We are looking for a Staff Perception Engineer to own our pedestrian detection model from data to deployment. You will be responsible for delivering a safety-critical perception system capable of detecting pedestrians at 100m or more within strict latency budgets on the Qualcomm 6490.

This is an end-to-end ownership role: modeling, training, evaluation, data strategy, deployment constraints, safety specification, and field-test methodology. You will collaborate closely with our embedded software team but retain full responsibility for the pedestrian detection roadmap.

What You’ll Do

• Own Saferide’s pedestrian detection model end to end: training, evaluation, data strategy, deployment performance, and continuous improvement.
• Define and maintain the perception quality bars, safety requirements, and field-testing methodology for pedestrian detection.
• Lead data collection, dataset curation, evaluation set design, and labeling pipeline for long-range pedestrian detection at 100m or more.
• Train, fine-tune, and optimize models using YOLOv11, RETDR, and DepthAnything frameworks.
• Profile and optimize model performance to meet strict edge-device constraints on the Qualcomm QCM6490 with a 50ms latency target.
• Diagnose failure modes, implement targeted data collection, and introduce architectural modifications when required.
• Collaborate with embedded engineers to integrate models into our on-device perception stack and debug performance issues.
• Own the safety specification for pedestrian detection, including test plans, metrics, and failure analysis.
• Drive the roadmap for incremental accuracy, latency, and robustness improvements across diverse real-world conditions.

What We’re Looking For

• 5 to 10 or more years of experience in machine learning, deep learning, or computer vision with significant work on detection models.
• Proven track record of shipping real-time perception systems on edge hardware such as mobile SoCs, embedded GPUs, NPUs, or DSPs.
• Strong experience with training and fine-tuning object detection models in the YOLO family, DETR variants, or transformer-based detectors.
• Experience with long-range pedestrian or vulnerable road user detection.
• Ability to design and manage datasets, labeling pipelines, and evaluation frameworks.
• Strong applied ML fundamentals including model debugging, augmentation strategy, optimization, and experimentation methodology.
• Experience optimizing models for quantization, pruning, ONNX, TensorRT, XNNPACK, QNN, or similar.
• Hands-on experience profiling latency and throughput on constrained hardware.
• Experience working with depth models or geometry-aware perception systems.
• Comfortable owning a safety-critical subsystem with demanding performance and reliability requirements.
• On-site in San Diego with willingness to participate in field tests and real-world data collection.

Bonus Skills

• Experience with autonomous driving, robotics, smart cameras, or ADAS perception.
• Familiarity with OC-SORT, ByteTrack, or other tracking frameworks.
• Experience with multi-camera fusion or depth-aware detection.
• Prior work on Qualcomm SoCs such as QCS or QCM series, Hexagon DSP or QNN, or Android NN frameworks.
• Prior ownership of safety-critical specifications or automotive validation processes including ISO 26262 or ASIL levels.
• Experience working with LFM or HDR automotive cameras.

Why Join Us

• Build a real safety-critical product that makes every vehicle on the road safer, not just expensive new ones.
• Own a perception system that will ship to millions of vehicles already on the road.
• Tight, elite engineering culture with deep hardware and embedded expertise.
• Hard technical problems including long-range vision, latency-constrained deployment, and safety-critical validation.
• Direct impact on a product that will be used by everyday drivers and families.

Our CEO previously scaled his first consumer electronics company to over $100M in revenue while shipping millions of units. Our CTO created 5G at Intel after serving as a staff engineer at Qualcomm for more than a decade. The rest of our engineering team are experts in computer vision, ADAS, and hardware systems. Together, we are building the platform that will bring life-saving safety technology to every vehicle on the road.

Compensation & Benefits

• Salary: $175,000 – $215,000 base, depending on experience and seniority.
• Equity: Meaningful stock in a rapidly growing venture-backed company. Your work directly builds company value, and we believe every team member should share in that upside.
• Benefits: health, dental, and vision coverage.
• On-site role in our San Diego office with direct access to hardware, vehicles, and test environments.