What You\’ll Do
Let’s build legal agents that actually think like lawyers.
We’re bringing AI agents into the real legal world – no gimmicks, no half-baked prototypes. With you on board, we’ll scale our working PoC into a secure, production-ready system that helps lawyers focus on what really matters.
You’ll own architecture, implementation, and ops – from fine-tuning and RAG to orchestration and RLHF. You choose the frameworks. You shape the stack. You build it end-to-end.
- Design intelligent agent workflows (planner/worker, swarms, etc.)
- Build and own the full AI pipeline: fine-tuning, RAG, evals, CI/CD
- Set up a scalable RLHF loop based on real legal feedback
- Run things reliably across Azure, AWS (or GCP if better)
- Implement solid security & GDPR-ready infrastructure
- Work hands-on with lawyers to test, iterate and ship
What you bring
- Strong Python skills and clean, production-grade engineering mindset
- Deep experience with LLMs, RLHF, fine-tuning, and RAG pipelines
- Hands-on with agent frameworks – or custom orchestration logic
- Familiar with eval & observability tools for LLM systems
- Solid cloud know-how (Azure, AWS or GCP) and IaC (Terraform, Pulumi, etc.)
- Bonus: legal NLP experience, explainable AI, German language skills
What we promise
- Build core infrastructure with full ownership and real impact
- Shape the agent tech stack – with budget to experiment and scale
- Work at the intersection of AI and law – in a real-world setting
- Office in Berlin-Mitte, high-end gear, hybrid setup (4:1)
- Smart, diverse, no-nonsense team – English first, German optional
- Ongoing learning: conferences, LegalTech roundtables, expert exchange
#J-18808-Ljbffr

Kontaktperson:
Legalhero HR Team