About Mariana Minerals
Mariana Minerals is a software-first, vertically integrated minerals company on a mission to supply the critical minerals powering modern energy, AI, and defense technologies. We’re reimagining the minerals supply chain by combining deep industry expertise with advanced software, automation, and data-driven decision-making.
The Role
Mariana Minerals is a software-first, vertically integrated minerals company supplying the minerals critical to modern energy, AI, and defense technologies. Our ML systems don't live in a vacuum — they see and act in our plants through sensors and robotic systems, they run as agents inside the tools the business uses every day, and they run on an ML platform that has to serve all of it.
We're hiring a Machine Learning Engineering Manager to lead the MLEs working on everything outside the chemistry and process models: perception and vision, sensor and robotics initiatives, LLM-powered and agentic workflows, and the ML platform and MLOps infrastructure the whole applied AI/ML organization depends on. You'll manage the people, own the technical quality of what ships, and partner with both Technical Product Managers — ML & Robotics, and Data & Analytics Platform — on what gets built and why.
This is a player-coach role with a wide surface area. The problems range from a camera on a conveyor to an agent answering an operator's question to the training and deployment infrastructure underneath both. Your job is to build a team of strong generalists, keep the platform coherent as the use cases multiply, and give the TPMs a counterpart who can say what is technically possible and what it will take.
What You'll Do
Manage, coach, and grow a team of ML engineers across perception, robotics, agentic workflows, and ML platform — hiring, onboarding, 1:1s, performance reviews, career development, and the hard conversations when they're needed.
Own the technical direction of the team: which approaches to bet on, how the platform is architected, and what "good enough to deploy" means for a vision model, a robot, or an agent.
Partner with TPMs on the roadmap: translate product priorities into scoped engineering work, push back when the ask isn't feasible, and commit to what the team will deliver.
Own the ML platform and MLOps stack as a product the rest of the ML org uses: training infrastructure, evaluation, deployment, monitoring, and the paved road that lets every MLE ship faster.
Own the engineering practices for perception and robotics: data collection and labeling, evaluation on real plant conditions, safety and fallback behavior, and the path from a demo to a system operators trust.
Own the engineering practices for LLM and agentic tools: evaluation, guardrails, cost, latency, and the discipline to know when an agent is the right answer and when conventional software is.
Own the seam with the software engineering organization at the engineering level — shared infrastructure, interfaces, and ownership — so nothing falls in the gap between the two orgs.
Own the production lifecycle of the team's systems — deployment, monitoring, incident response, on-call — and the reliability bar for systems that act in a plant.
Stay hands-on enough to review the hardest work, unblock engineers, and prototype when the fastest path to an answer is to build it yourself.
Own headcount planning and hiring for the team, and build a pipeline of ML engineers who can move between problem areas.
How You'll Operate
Player-coach: You spend most of your time on people, priorities, and quality — and enough time in the code and the systems to keep your judgment sharp.
Structure from ambiguity: Turn a loosely defined plant problem or internal-tool idea into scoped engineering work with clear success criteria, and align operators, MLEs, software engineers, and product behind it.
Breadth with depth on demand: You're credible across perception, LLMs, and infrastructure, and you know when a problem needs a specialist.
Clear commitments: Give the TPMs and the business honest estimates, visible tradeoffs, and early warning when something is slipping.
Ecosystem fluency: Understand the Mariana ML, perception, robotics, and data ecosystem as a whole — and where it sits relative to the current frontier of model capability — so the team's bets stay well-calibrated.
What We're Looking For
Must have
6+ years in machine learning engineering, including 2+ years managing ML engineers with direct responsibility for hiring, performance, and growth.
Hands-on track record shipping ML systems to production in at least two of: computer vision / perception, robotics or embedded ML, LLM-powered or agentic applications, ML platform / MLOps infrastructure.
Strong software engineering fundamentals: you've built and operated production systems, you care about interfaces and reliability, and you can hold a design review with software engineers as a peer.
Working fluency with the current LLM/foundation-model landscape — what the models can do today, how to evaluate them, and where agentic approaches genuinely fit versus conventional software.
Track record of setting technical direction for a team and delivering against commitments in an ambiguous, cross-functional environment.
Ability to evaluate work you didn't do — you can review a model, a pipeline, or a system design and give feedback that makes the engineer better.
Exceptional written and verbal communication across audiences, from ML engineers to operators to executives.
Nice to have
Prior work on industrial robotics, perception in harsh environments, or closed-loop control of physical systems.
Experience owning an ML platform or MLOps stack used by multiple teams.
Experience shipping LLM-powered or agentic internal tools to real users and measuring adoption.
Background in mining, energy, chemicals, manufacturing, or other heavy industry — especially industrial automation or sensor/vision data.
Experience building a team from a handful of engineers to a functioning organization.
Why This Role
Most ML manager roles hand you one problem area and a platform someone else owns. This role hands you the perception systems that will see our plants, the robotic systems that will act in them, the agents that will run inside our tools, and the platform underneath all of it — at the moment they move from ad hoc experimentation into supported systems. At Mariana, you don't need to validate product-market fit, because we are the market: if your team builds something that works, it gets deployed.
Our culture is built on four principles:
Everyone Gets Home Safe. We never put speed or cost ahead of people.
Extreme Ownership. We take full responsibility for outcomes, relentlessly driving toward solutions.
Engineer Out Requirements, then Automate. We simplify, optimize, and then automate for scale.
Share Your Legos. We collaborate openly, share knowledge, and empower each other to build bigger, better solutions.
Join us as we build the future of responsible mineral sourcing and supply!