The role
- COMP
- $150K - $250K
- EQUITY
- Competitive equity
- LOCATION
- San Francisco
- WORKPLACE
- Hybrid
- EXPERIENCE
- 3+ years
- VISA
- None
- STACK
- Python, PyTorch, Vector Database
- INDUSTRY
- Software Development, Hardware, Defense, B2B
/roles — ROLE_384
Founding Machine Learning Engineer
Pre-seed startup automating the scoring of engineering skill assessments for space, defense, and robotics employers
The company
An early-stage hiring platform for hardware and deep-tech teams in fields like aerospace, automotive, energy, and defense, using AI simulations of real engineering work to assess candidates.
- STAGE
- seed-stage
- FUNDING
- seed funding
- TEAM
- ~10 people
- FOUNDED
- 2025
JD — the work
About the role
Leading companies in robotics, manufacturing, space, and defense use this platform's assessments to choose their next engineers, and turning that scoring into an automated system is an open ML problem you would own. Deterministic checks, such as whether a completed design is within spec, already work; the unsolved part is judging the reasoning path a candidate took, including options considered and rejected. You would report to the co-founder and CTO and partner with in-house mechanical and electrical engineering experts who set the standard for a correct answer.
What you'll do
- Build multimodal scoring models that read circuit board layouts, 3D CAD, simulation results, and logs of candidate actions
- Turn expert engineering judgment about process, not just outcomes, into machine-readable scoring rubrics
- Run the full training cycle: generate data from live assessments, capture trajectories, analyze failures, curate datasets, and evaluate continuously
- Work side by side with domain experts to convert their assessments of candidates into structured training labels
- Ship models into production scoring pipelines that inform live hiring decisions
What they're looking for
- 3+ years of machine learning engineering experience
- Strong Python and PyTorch skills for production models
- Ability to own an open-ended ML problem from data to deployment
- Comfort working closely with non-ML domain experts to define labels and rubrics
- Willingness to work hybrid in San Francisco
Nice to have
- Prior hands-on experience with vector databases