Data Scientist - Geostatistics, Geophysics

Looking for a self-driven data modeller to join a venture-backed mining technology start-up that is transforming how minerals are mapped and extracted.

Why this matters

To meet global demand, the world needs to mine as much copper in the next 25 years as in all of human history. Today's methods cannot scale to this challenge. We are developing frontier technology to fix this, starting with the core of every mining company: the understanding of the orebody.

We're a Cambridge University spin-out using explainable AI and value-of-information analysis to transform how mines are developed and operated. Our uncertainty-aware models map subsurface resources from multimodal data, so mining companies can make faster evidence-based decisions on where to drill next and how best to move sites forward. The outcome is a halving of drilling costs, accelerated time to production for new projects, and reduced project risk for operational mines so more minerals make it to market.

We recently closed an oversubscribed $4m round, have POCs running with some of the largest mining companies in the world and now have more pilot demand than we can serve. We're expanding the team to turn breakthrough technology into a product and scale it.

The Role

Three things, done well, every week:

• Turn models into decisions. The ML team hands you probabilistic block-model outputs and drill planning optimisation options. The geologist hands you a client’s requirements. You shape the model into an output that's right for this deposit and ready for a geologist to explain to the client.

• Benchmark against the standards. You benchmark our approach head-to-head against the methods clients already use (e.g. ordinary and indicator kriging, conditional simulation) so we can communicate the exact benefits and honest trade-offs.

• Build the foundation, not the notebook. You leave behind reusable, tested workflows the next deposit inherits, not a one-off script only you can run.

And one bigger thing, over months and years:

• Shape the methodology. You'll have real influence over how we model the subsurface and how we quantify and communicate uncertainty. If you spot a better way to do something, you'll be backed to pursue it.

Who You Are

• Mission-driven. There are easier, better-paid jobs elsewhere. Consider whether building a company that transforms a foundational industry is your main reason for applying.

• Self-directing. You identify what needs doing without being told, and prioritise ruthlessly.

• Strong mining experience. Real experience using geological and geophysics data to guide decisions in mining.

• Strong geostatistics. Variography, Kriging variants, conditional simulation, SGS etc… are familiar concepts that you can implement it, not just name.

• Fluent in Python. You can take someone else's model outputs and reshape them into something rigorous and reproducible.

• A sharp communicator, written and verbal. You can hand a resource geologist a result they'll recognise and defend, and hold your ground with senior geologists without overplaying your hand.

• Reach for uncertainty by instinct. You think in confidence, not point estimates, and you're comfortable explaining the intricacies of evolving distributions with new information.

• Unsatisfied and prone to action: it unsettles you when things could be done better and when you’ve identified these in previous roles you’ve implemented something that changed the organisation.

• Unreasonably resilient. You're proud of having done something genuinely hard; the challenge was a large part of the motivator.

Bonus if you:

• PhD or Masters in Geophysics or Geostatistics

• Have contributed to JORC or NI 43-101 compliant mineral resource estimates, or worked alongside Competent/Qualified Persons.

• Are fluent in Leapfrog, Vulcan, Datamine, Micromine.

• Can reason about what an ML model is doing under the hood. You’ll work closely with the ML development team.

• Have worked across multiple deposit styles and commodities

• Have opinions about what's wrong with resource modelling and mine planning today.

How We Think About Ore Body Knowledge

We treat OBK as one continuous discipline, not a relay race between explorers, resource geologists and mine geologists. We combine drillhole data, structural interpretation, geophysics and geochemical data to give teams across that span a faster, more reliable quantification of the properties of their deposit.

Human-in-the-loop throughout: we combine OBK modelling with putting value-of-information analysis in the hands of users so that geologists can use their judgement to come to the best decisions for their projects. We expect the workflow to keep evolving, and we want someone to help evolve it.

How you will work

We're a small team with a lot of momentum and there is no playbook yet. You'll often have to decide what the right thing to do is, and not just execute on tasks. Some weeks are ambiguous and occasionally on fire. If that energises you, you'll thrive here; if you need a defined backlog handed to you, you won't.

• King's Cross office — 4 days/week in person

• Meaningful equity

The Team

• Small & technical - 4 people, all engineers by background.

• Advisor support - you will be supported by industry veterans including Matt Mullins.

• Gabriel Yoong - CEO and Co-Founder

• Luke Cullen - CTO and Co-Founder

Apply or Refer

If you're interested, or know someone who is, we'd like to hear from you.
Email your CV to gabriel.yoong@materialdifference.earth and we will get back to you.