PyTorch
ABOUT THE ROLE
Join an early-stage industrial AI team building wearable AI copilots for field workers. You will own core parts of the multimodal AI stack, from production vision-language systems to real-time experiences running on hardware in the field.
WHAT YOU'LL DO
- Build and ship agentic vision-language pipelines for smart glasses, including multi-step visual reasoning and tool use in industrial workflows.
- Develop evaluation harnesses, capture failure modes, and turn customer data into fine-tuning cycles that improve model quality.
- Adapt model orchestration and edge inference to variable connectivity, latency, and hardware constraints.
- Create real-time voice and video interfaces, including conversational and proactive alert experiences.
- Fine-tune and optimize open-weight multimodal models for enterprise and on-premise deployments.
WHAT WE'RE LOOKING FOR
- Two to six years of experience shipping production vision-language or multimodal AI systems, using Python and PyTorch.
- End-to-end ownership of production AI products, including model orchestration, evaluation, and downstream product outcomes.
- Experience building multi-step agents with real tool use, rather than only retrieval wrappers or single-turn chatbots.
- Applied expertise in fine-tuning methods such as SFT and RLHF, evaluation design, and visual reasoning; pretraining from scratch is not the focus.
- Familiarity with technologies such as vLLM, Triton, Ray Serve, ONNX, TensorRT, Hugging Face Transformers, quantization, and edge inference.
- Startup, founding-engineer, or ot