Senior Full-Stack AI Engineer – DeepLaw
DeepLaw is building an intelligent AI agent that reviews contracts — not just for legal risk, but through a business lens . Our goal is to help founders, operators, and legal teams understand what a contract actually means for them: the assumptions it relies on, what trade-offs it reflects, and how it aligns (or conflicts) with their values and goals.
We're looking for a Senior Engineer to lead technical development from scratch. You’ll build the agent system, connect the pieces (UI ↔ LLM ↔ data layer), and work closely with the founding PM to define the product and roadmap.
🛠 What You’ll Build
-
The core AI agent pipeline — likely involving prompt orchestration, retrieval (RAG), and fine-tuning
-
UX and tooling for interacting with the agent (basic web UI, annotation, structured contract data)
-
Data + model infrastructure to support iteration and experimentation
-
MVP features: assumption mapping, contract diagnostics, value-aligned feedback, etc.
-
Deployment setup for a working prototype (CI/CD, cloud infra, minimal ops)
-
Help define what “value-aligned” contract review actually means in product and code
🧩 Who You Are
-
5+ years of experience as a full-stack engineer or AI engineer, ideally at startups
-
Experience with LLMs in production: prompt engineering, LangChain, AutoGen, or similar
-
Familiar with agent-based architectures and RAG pipelines
-
Comfortable spinning up web apps, APIs, vector stores, and fine-tuning loops
-
Bonus: experience with legal tech, working with contracts, or interest in reasoning-focused AI
-
Collaborative, autonomous, and energized by zero-to-one product challenges
🧠 Why This Role Is Unique
-
Work on a real-world application of LLMs that goes beyond summarization
-
Partner directly with a product-minded founder from day one
-
High autonomy, fast decisions, greenfield architecture
-
Influence not just what we build — but how we work and who we hire next
🔧 You’ll Probably Touch…
- OpenAI, Claude, open-source models
-
LangChain / AutoGen / or a roll-your-own agent layer
- Pinecone, Weaviate, Postgres, Supabase
- React/Next.js or equivalent for UI
-
GCP, AWS, Vercel, or lightweight infra for fast iteration
Originally posted on Himalayas