Job title: Founding AI Engineer
Job type: Permanent
Emp type: Full-time
Industry: AI & Hardware
Salary type: Annual
Salary: Negotiable
Job published: 16/11/2026
Job ID: 299859

Job Description

Founding AI Engineer - Reinforcement Learning

Location: Cambridge or London (hybrid)

Salary: £85,000 to £100,000 + EMI equity

 

About the company

We’re a VC backed AI security startup building tools that red team autonomous AI agents. As companies roll out agents that read email, call tools, connect to MCP servers and write code, they’re opening up attack surfaces nobody has properly tested. We build the models and agents that find those weaknesses first.

 

We’ve raised £2m and we’re a team of three. A security advisor to the UK Government sits on our advisory board. We plan to raise our Series A in the next 9 to 12 months.

 

The role

This is our first engineering hire. You’ll build the core AI IP behind the product. That means training our own models, deploying them, and turning them into agents that attack other AI systems. You’ll work directly with the CTO and have real ownership of technical direction from day one.

 

What you’ll be doing

  • Training and post training open source models in the 7 to 8B parameter range
  • Running RL based post training, including GRPO, and designing reward functions that produce the behaviour we need
  • Fine tuning models for offensive security tasks and benchmarking them over time
  • Taking models from Hugging Face through to deployment, hosted across multiple cloud providers and served via key based APIs
  • Building red teaming agents that probe and attack autonomous systems such as email agents, MCP servers, LLM applications and coding agents
  • Testing and comparing our models against frontier models from OpenAI, Anthropic and others
  • Building out the training and evaluation pipeline that becomes core company IP

 

What we’re looking for

  • Hands on experience fine tuning and post training LLMs, ideally with RL methods such as GRPO, PPO or DPO
  • A strong understanding of reward design and model evaluation
  • Experience deploying and serving models in production, not just training them
  • A security mindset. You think about how AI systems can be broken, not just how they’re built
  • Comfortable in an early stage environment where you build from scratch and own the outcome
  • A PhD is welcome but not essential. Depth of hands on experience counts just as much

 

Nice to have

  • Background in AI red teaming, adversarial ML or vulnerability research
  • Experience with agent frameworks, tool use or MCP
  • Experience at a research lab or security team working on AI safety or model robustness

 

What we offer

  • £60,000 to £100,000 base salary depending on experience
  • Meaningful equity through an EMI scheme
  • A founding role with direct influence over the product and the company’s technical direction
  • The chance to work on one of the most important open problems in AI right now