Edge ML Application Developer

This engagement focuses on building the critical integration layer between the client’s on-vehicle machine learning models and real-time, driver-facing features on an edge-compute platform. The internal science team owns core model development; our role is to partner with them to deploy, optimize, and productionize those models on-device; translating research-grade models into reliable, real-time application logic. This includes preparing and synchronizing sensor inputs, tuning models for on-device performance constraints, and architecting the logic that arbitrates and combines outputs from multiple concurrent models into a single, dependable feature decision the driver can trust., * Port, compile, and deploy ML models to resource-constrained edge-compute platforms.

• Optimize model inference for latency, memory, throughput, and hardware constraints.
• Build preprocessing pipelines for camera, telematics, and other sensor inputs.
• Develop post-processing and application logic that combines outputs from multiple concurrent models into unified real-time decisions.
• Implement confidence filtering, prioritization, and arbitration logic across competing model outputs and driver notifications.
• Integrate perception outputs into real-time vehicle features such as alerts, visual indicators, or audible warnings.
• Collaborate with perception, platform, embedded, and OS engineering teams to ensure sensor-data, timing, and runtime compatibility.
• Execute against an established system architecture while iterating quickly as requirements and implementation details evolve.

• Strong Python experience with hands-on deployment and optimization of machine learning models on edge or embedded devices.
• Experience with edge inference frameworks such as TensorRT, ONNX Runtime, TensorFlow Lite, or comparable technologies.
• Experience deploying production computer vision, perception, or deep-learning models in real-time or near-real-time environments.
• Prior experience within automotive, ADAS, autonomous vehicles, robotics, drones, or another sensor-driven autonomous system.
• Experience working with camera, LiDAR, radar, IMU, or other sensor-based perception inputs.
• Experience integrating outputs from multiple models or perception components into unified application or decision logic.
• Strong understanding of latency, memory, throughput, and compute constraints in edge environments.
• Experience with preprocessing, post-processing, confidence thresholds, filtering, tracking, fusion, or output arbitration.
• Ability to work North American business hours with strong written and verbal communication skills.

Bonus Experience

• Experience with NVIDIA Jetson, Orin, DRIVE, CUDA, or DeepStream.
• Experience with model quantization and optimization techniques such as INT8, FP16, pruning, distillation, or layer fusion.
• Experience with ADAS, collision avoidance, driver monitoring, or other safety-critical vehicle systems.
• Experience with sensor fusion, multi-camera perception, LiDAR processing, trajectory estimation, or 3D perception.
• Strong C++ experience for performance-sensitive inference or perception applications.