Job description

San Francisco, CA · On-site · Full-time Compensation: $180,000–$230,000 + 0.25–0.75% equity About the Company Our client is an early-stage, venture-backed startup building wearable AI assistants for industrial field workers: smart glasses powered by agentic vision-language models. Their product is deployed in critical-infrastructure and heavy-industry settings where AI has rarely been applied, and they are already working with anchor enterprise customers alongside a strong pipeline of large enterprise prospects. Recently founded · Founding team stage · Industry: AI, Hardware, Robotics, Enterprise, B2B The Role Our client is hiring a founding machine learning engineer to own the AI core of their product from day one. You will build and ship multimodal systems that run on real hardware in the field, working directly with the founding team in a fast-moving, in-person environment. This role suits a hands-on builder who has put vision-language or multimodal systems in front of real users. What you'll be doing • Building a production pipeline of multi-step, tool-using visual reasoning agents that run on smart glasses against real industrial workflows such as inspections and standard operating procedures • Creating real-time voice and video AI interfaces for the glasses, including conversational and proactive alert modes tailored to different users • Owning evaluation and the data flywheel: eval harnesses, capturing failure modes, and turning customer data into fine-tuning loops that improve model quality release over release • Delivering edge inference and model orchestration that adapts gracefully to changing connectivity and latency constraints in the field • Fine-tuning and optimizing open-source multimodal models (SFT, RLHF, quantization) for on-premise enterprise deployments Tech stack: Python, PyTorch, vLLM, Triton, Ray Serve, ONNX, TensorRT, Hugging Face Transformers, LangChain, RAG, RLHF, SFT, quantization (GPTQ, AWQ), edge AI, multimodal LLMs, vision-language models, Docker Requirements • Up to 3 years of hands-on experience building and deploying multimodal or vision-language AI systems in Python and PyTorch • A track record of shipping vision-language systems that real users rely on in production, where you owned both the model and the orchestration layer, rather than demos or research prototypes • Practical depth in applied model work such as fine-tuning (SFT, RLHF), evaluation design, and orchestrating models in production, including visual reasoning and detection or segmentation where needed • Experience building agents that plan across multiple steps and call tools, along with the evaluation harnesses and data loops that keep them improving • Experience as a founder or very early engineer at a startup, or at a fast-paced, high-intensity engineering organization • A bachelor's or master's degree in computer science, machine learning, or engineering from a strong program, or equivalent experience shown through production AI work • Real enthusiasm for computer vision, wearables, and industrial AI, visible in your projects, side work, or career path • Ability to work on-site in San Francisco five days a week; openness to shared team housing is a plus • Existing US work authorization; our client can support visa transfers (for example OPT or H-1B transfer) but cannot sponsor new visas Nice to Haves • Experience shipping AI for AR or wearable devices, or computer vision for autonomous driving • A master's degree that included vision or multimodal research, such as a thesis or published work • Experience deploying open-weight models at the edge or on-premise on constrained hardware using tools like vLLM, Triton, TensorRT, or quantization techniques Why Join • Work on one of the more interesting applied computer vision problems in the market, running on real hardware in real industrial environments • Own the full AI stack as a foundational member of the engineering team • Join a company with live enterprise customers and meaningful commercial momentum • Meaningful founding equity alongside a competitive base salary • A high-ownership, in-person culture built around shipping fast Details • Location: San Francisco, CA • Work policy: On-site 5 days per week • Compensation: $180,000–$230,000 + equity • Visa sponsorship: No new sponsorship; open to visa transfers (for example OPT or H-1B transfer) • Employment type: Full-time

Requirements

Must-haves

San Francisco, CA · On-site · Full-time
Compensation: $180,000–$230,000 + 0.25–0.75% equity
What you'll be doing
Building a production pipeline of multi-step, tool-using visual reasoning agents that run on smart glasses against real industrial workflows such as inspections and standard operating procedures
Creating real-time voice and video AI interfaces for the glasses, including conversational and proactive alert modes tailored to different users
Owning evaluation and the data flywheel: eval harnesses, capturing failure modes, and turning customer data into fine-tuning loops that improve model quality release over release
Delivering edge inference and model orchestration that adapts gracefully to changing connectivity and latency constraints in the field
Fine-tuning and optimizing open-source multimodal models (SFT, RLHF, quantization) for on-premise enterprise deployments
Tech stack: Python, PyTorch, vLLM, Triton, Ray Serve, ONNX, TensorRT, Hugging Face Transformers, LangChain, RAG, RLHF, SFT, quantization (GPTQ, AWQ), edge AI, multimodal LLMs, vision-language models, Docker
Up to 3 years of hands-on experience building and deploying multimodal or vision-language AI systems in Python and PyTorch
A track record of shipping vision-language systems that real users rely on in production, where you owned both the model and the orchestration layer, rather than demos or research prototypes
Practical depth in applied model work such as fine-tuning (SFT, RLHF), evaluation design, and orchestrating models in production, including visual reasoning and detection or segmentation where needed

Nice-to-haves

Ability to work on-site in San Francisco five days a week; openness to shared team housing is a plus
Experience shipping AI for AR or wearable devices, or computer vision for autonomous driving
A master's degree that included vision or multimodal research, such as a thesis or published work
Experience deploying open-weight models at the edge or on-premise on constrained hardware using tools like vLLM, Triton, TensorRT, or quantization techniques
Work on one of the more interesting applied computer vision problems in the market, running on real hardware in real industrial environments
Own the full AI stack as a foundational member of the engineering team
Join a company with live enterprise customers and meaningful commercial momentum
Meaningful founding equity alongside a competitive base salary
A high-ownership, in-person culture built around shipping fast
Location: San Francisco, CA
Work policy: On-site 5 days per week
Compensation: $180,000–$230,000 + equity

Application prep

Use the role’s language and application details while tailoring your materials.

Keywords to mirror

aimachine learningpythondocker

Platform: manatal
Effort: medium

Benefits

Benefits detected in the listing.

Vision