AI Research Engineer – Vision AI, VLM, Physical AI

This is a fully remote position, open to applicants in Washington.

📋 Description

• Develop and enhance models for detection, tracking, segmentation, pose and activity recognition, and scene comprehension.

• Train and assess vision-language models for grounding, dense captioning, temporal question answering, and tool utilization.

• Design retrieval-augmented and agentic loops for tasks involving perception and action.

• Prototype perception-in-the-loop policies utilizing both simulated and real-world data.

• Integrate systems with planners and task graphs for workflows related to manipulation, navigation, and safety.

• Curate datasets, establish evaluation protocols and key performance indicators (KPIs), and conduct ablation studies.

• Package research findings into dependable services using technologies such as Kubernetes, Docker, Ray, and FastAPI.

• Implement profiling, telemetry, and continuous integration (CI) for reproducible scientific outcomes.

• Manage multi-agent pipelines that combine perception, reasoning, simulation, and code generation.

• Convert research from theoretical papers to prototypes and ultimately to deployable production modules.

• Produce publishable or open-source results, or production-ready modules that enhance product KPIs.

• Generate reproducible code, evaluation reports, and capability demonstrations.

⛳️ Requirements

• Master’s or Ph.D. in Computer Science, Electrical Engineering, Robotics, or a related field.

• Actively publishing in computer vision, machine learning, or robotics forums such as CVPR, ICCV, ECCV, NeurIPS, ICML, ICLR, CoRL, or RSS.

• Proficiency in PyTorch or JAX and Python programming.

• Experience with CUDA profiling and mixed-precision training.

• Proven research experience in computer vision, with expertise in at least one area such as vision-language models (VLMs), embodied or physical AI, or 3D perception.

• Capability to transition from theoretical papers to coding, conducting ablations, and yielding results with meticulous experiment tracking.

• Preferred: experience with video models, diffusion or 3D generative synthesis/NeRF pipelines, or SLAM/scene reconstruction.

• Preferred: experience in multimodal grounding or temporal reasoning.

• Preferred: familiarity with ROS2, DeepStream/TAO, TensorRT, or ONNX.

• Preferred: experience with Ray and distributed data loaders or sharded checkpoints.

• Preferred: knowledge of testing, linting, profiling, containers, and reproducibility practices.

• Preferred: public GitHub code contributions and first-author publications, or significant open-source impact.

🏝️ Benefits

• Make a real impact: research is implemented, driving core features in MVPs and products.

• Receive mentorship from a Principal Architect and senior engineers/researchers.

• Enjoy a balance of leading research methodologies with a pragmatic focus on product development.