Founding Machine Learning Engineer

About the company

Our client is an early-stage startup whose AI simulations of real CAD work help space, defense and robotics teams hire engineers. The team is around ten people, already holds multi-year enterprise contracts, and is working on generative CAD — getting models to reason about mechanical and electrical design the way an experienced engineer does. The data they train on is proprietary and comes from real engineering workflows rather than scraped from the public internet.

The role

This is the founding machine learning hire, and the mandate is a genuinely unsolved problem: owning the development of a novel engineering scoring system that sits at the core of the company's intellectual property. You would work directly with the mechanical and electrical engineering leadership, training models on proprietary data captured from real-world engineering workflows, and help define what generative CAD looks like by mapping the full landscape of possible design paths. It is foundational ML work — training models and owning the data pipelines underneath them — not application-layer work on top of someone else's foundation model.

What you'll do

• Own the development of a novel engineering scoring system end to end — the core piece of intellectual property, and a problem the industry has not solved
• Train models and design architectures against proprietary data captured from real-world mechanical and electrical engineering workflows
• Build and own the multimodal data pipelines behind them — CAD geometry, simulation output, video and interaction traces
• Work directly with the mechanical and electrical engineering leadership to turn domain judgement into something a model can learn
• Help map the landscape of possible design paths that generative CAD has to reason across
• Operate as a founding hire on a ten-person team with established multi-year enterprise contracts — high ownership, startup cadence, occasional urgent surges

Nice to have

• Machine learning experience on a CAD or generative-design AI team — the closest possible match for this work
• ML experience in chips and EDA, or another hard-tech engineering discipline
• Reinforcement learning or RLHF exposure — representation and geometry learning transfer well here, though it is not a requirement
• Experience turning expert human judgement into a scoring or evaluation model
• Already based in the San Francisco Bay Area

Tech stack

Foundational machine learning — model training, architecture design and multimodal data pipelines over CAD geometry, simulation output and interaction traces. Reinforcement learning and geometry or representation learning are directly relevant.

Compensation & logistics

• $150k–$250k base plus competitive equity — this is the founding machine learning seat
• Full-time and hybrid in Mission Bay, San Francisco — minimum 3 days per week in the office, with a preference for 5; relocation is expected for candidates who are not local
• No visa sponsorship of any kind — candidates must be legally authorized to work in the United States without visa sponsorship, now or in the future. TN status works only if you arrange the TN yourself; the client files nothing and does not sponsor H-1Bs
• 3–8 years of machine learning engineering experience — the 3-year floor is a hard gate, the upper end is flexible
• Startup cadence: a 5-day working week as standard, with occasional urgent surges, and founding-hire ownership expectations
• One hire