Senior Machine Learning Engineer

Most ML roles sit in one of two worlds: research teams publishing papers, or Big Tech teams maintaining infrastructure for clean data 're neither.

Propel Labs is an IDEO-style studio for AI and tough tech
, focused on mining technology. We're building AI-powered tools that help geoscientists make better decisions about what's underground—precision extraction, ore body modeling, uncertainty quantification. The kind of problems where getting it wrong costs millions and getting it right changes the industry.

We have something rare:
direct access to world-class geoscientists through our operating partners. You won't be guessing what users need. You'll sit in rooms with domain experts, translate their workflows into ML systems, and ship features they use the same week.

This is a founding ML role with real ownership. You'll set the technical foundations, make architectural decisions that stick, and have equity in a company at the frontier of mining technology.

• Take requirements from geoscientists → ship working ML features within weeks, not quarters
• Build data pipelines from raw geological data → model-ready features
• Implement uncertainty quantification that geoscientists will actually trust
• Deploy models that work in production, not just notebooks

• Sit in meetings with domain experts, ask the right technical questions
• Translate "we need variogram auto-fitting" into a working system
• Validate model outputs with domain experts before shipping

• Make pragmatic build vs. buy decisions (fine-tuning vs. off-the-shelf)
• Design for iteration speed, not perfection

Within your first 90 days, you will ship a working ML feature that geoscientists at our partner company use in their daily workflow.

This isn't a probation period—it's the pace we operate at. If you're someone who needs 6 months to ramp up before contributing, this isn't the right fit. If you thrive on shipping fast with real users, you'll love it here.

• 5-8 years total, with at least 3 years in production ML (not just research)
• Background in applied ML for scientific or industrial domains: geospatial, climate, energy, manufacturing, robotics
• Track record of ML systems that non-technical domain experts actually use and trust

• Python fluency, PyTorch or Tensor Flow
• Comfortable with numerical computing and messy real-world data
• Bonus: experience with spatial data, time-series, or uncertainty quantification

• Pragmatist who ships over perfectionist who polishes
• Comfortable with ambiguity; makes progress when requirements are fuzzy
• Genuinely curious about domains outside ML; wants to learn how geoscientists think

We want to be honest about fit. This role isn't right for:

• Pure researchers who want to publish papers before shipping products
• Big Tech ML engineers who need massive infrastructure, clean data, and clear requirements
• Candidates who can't work directly with non-technical domain experts
• People who need certainty before they start building

• Direct domain access:
  You'll work alongside geoscientists who can validate your work in real‑time
• Ship fast:
  Features go from idea to production in weeks, not quarters
• Real impact:
  Mining tech affects global resource extraction, sustainability, and ESG outcomes
• Founding role:
  Set the ML foundations and own the technical direction
• Equity:
  Meaningful ownership in a pre-Series A company

We respect your time. Our process is fast and transparent:

• 30 min Screening Call for culture fit, motivation, high-level experience
• 60 minutes Technical Deep Dive - Past ML systems you've shipped, technical decisions
• 90 Minute Domain Challenge - Design an ML system for a domain you don't know
• Teem meet & offer - Meet the team, finalize details

Send your resume and a brief note on an ML system you shipped to:
careers

• No cover letter required. We'd rather see your work.