Applied AI Role | Agentic Systems for Legal Workflows
Remote (GMT-5 to GMT+1 overlap) | Optional hubs: London / New York
Full-time | Mid-level
£100,000 – £200,000 + meaningful equity
Our client — AI-native LegalTech company
Travel: London every 5–6 weeks
Work Authorization
About the Opportunity
Our client is rebuilding legal services around AI and data. Backed by $9M+ from a tier-1 US VC, they are building a Cursor-like experience where lawyers collaborate with intelligent agents on real legal work.
You will join a small, high-caliber engineering team growing from 2 to 5, with direct ownership over agent and retrieval systems used daily by internal lawyers.
This is an engineering-first applied AI role focused on shipping reliable, production-grade systems — not just experimentation.
What You'll Build
Retrieval pipelines that ground outputs with high-quality citations
Deep research workflows enabling step-by-step legal investigation
Model fine-tuning pipelines as internal data matures
End-to-end AI features deployed into real production environments
Tech Environment
Python (core language)
LLM-powered and agentic architectures
Retrieval and search systems
Production backend infrastructure
AI-assisted engineering workflows
What They're Looking For
Strong production experience building AI systems in Python
Hands-on work with LLM-powered or agentic systems
Proven model training or fine-tuning experience
Ability to translate ambiguous product goals into working systems
Track record of shipping end-to-end in fast environments
Nice to Have
Experience with vector search or hybrid retrieval
Background in collaborative editor or document workflows
Exposure to regulated or high-stakes domains
Familiarity with GCP, Postgres, or Temporal
Experience building products from zero to one
Why This Role Stands Out
Meaningful equity in a well-funded early-stage company
Unlimited budget for AI coding tools
Real ownership of agentic infrastructure
Annual Mallorca offsite + regular team gatherings
Tight feedback loop with real legal users
If you have shipped real AI systems into production and care about making agents actually reliable, our client wants to meet you...