What is Overmind:

AI agents are going into production at banks, hospitals, and governments. The problem? Existing security tools weren't built for systems that make autonomous decisions in real-time. We're building Overmind - the security layer that actually understands how AI agents work. We learn each agent's behaviour patterns and catch the subtle failures that matter, without breaking production workflows.

This is infrastructure-level work at the frontier of AI deployment. We're defining how autonomous systems operate in environments where mistakes have real consequences. The opportunity

We're not just securing AI - we're unlocking its potential. As a Founding ML Scientist you will:

• Build security that evolves: Help build the next generation of AI security that learns and adapts to each agent it protects. As AI systems get more sophisticated, so do the defences.

• Define infrastructure for a new era: You're not building another monitoring tool. You're establishing the control layer for AI making real-time decisions in banks, hospitals, and defence.

-Shape what we build and why: You're a founding team member, not just a hire. Help define the problem space, shape the product vision, and decide what AI security needs to be. Your technical insights drive company strategy.

• Own your domain completely: Full autonomy over architecture, tooling, and standards. You decide how to build and scale. No approval chains. You ship, iterate, and help set the technical culture.

The team: You'll be working directly with the founding team:

• Tyler Edwards (CEO): Former Intelligence Officer working across MI5/MI6/GCHQ. 3 AI safety publications. Built tech for Defence and intelligence operations.

• Akhat Rakishev (CTO): Former Data Scientist and ML Engineer from Monzo and Lyst. Architected and scaled ML pipelines, led ML projects adding $1M+ ARR.

• Sam Brunt (CRO): GTM leader with 8+ years at B2B startups. Scaled Vertice from traction to $5-20M ARR. First GTM hire EMEA at Pipe.

We didn't stumble into this problem. We've lived it across intelligence, fintech, and infrastructure software. Overmind is the product of hands-on experience at the sharpest edge of deployment.

What we're looking for

Technical depth: -Training LLMs with modern methods e.g. RL with GRPO, fine-tuning with QLoRA etc. -Setting up multi-GPU training and inference infrastructure in cloud (GCP/AWS) -Using modern optimisation libraries for efficient training and inference e.g. vLLM, Unsloth, torchtune

Founding engineer mindset: -You've built things from scratch before, whether at a startup, on side projects -You're comfortable with ambiguity and can make pragmatic technical decisions with incomplete information -You default to shipping and iterating rather than over-engineering

Bonus points for: -Experience in regulated industries (fintech, health-tech, defence) -Background in AI safety, security, or evaluation -Open-source contributions or projects in the AI/ML space -Previous founding or very early-stage startup experience

Ways of working -We build our team with these values in mind: -Default to trust. We hire exceptional people and get out of their way. Autonomy over process. -Pursue impact. We're obsessed with outcomes. Your work needs to create real value - that's what matters. -Work intentionally. Deep focus when you need it. High-bandwidth collaboration when it matters. Flexibility is key - we care about outcomes, not where you're sitting.