Job Description & Details
This role puts you squarely in the hot seat as a solo engineering force embedding AI agents directly into JLL's legal systems. You will bridge the gap between central AI architecture and business-line reality, building high-leverage workflows where you actually own the production outcome rather than just writing prompt wrappers. If you enjoy operating autonomously and shipping real-world tools that lawyers will actively use, this is a solid playground.
What You'll Actually Be Doing
You will spend your days shoulder-to-shoulder with legal subject matter experts, turning their manual bottlenecks into automated agentic workflows and Model Context Protocol (MCP) servers. Expect to stand up connectors, bootstrap evaluations, run internal workshops on tools like Claude Code, and eventually hand mature systems off to operations once they stabilize. The real challenge here is balancing rapid prototyping with enterprise-grade reliability since you are effectively acting as a one-person engineering startup inside a massive corporate legal department.
The Core Tech Stack
You need to be completely fluent in Python, Node, or .NET while being dangerous enough to jump across all three when an integration demands it. Experience with the Claude Agent SDK, MCP toolchains, shared RAG services, and LLM gateways is non-negotiable since you'll be building directly on top of them. On top of the AI layer, you'll need deep fluency across relational, NoSQL, vector, and document databases because the hardest part of this job won't be calling the LLM—it'll be getting enterprise data structured and secured properly.
Interview Expectations
The technical round is going to lean heavily into real-world systems architecture and failure modes rather than LeetCode puzzles. They will likely ask you to design an agentic workflow that handles messy, unstructured legal documents while gracefully recovering from hallucinated tool calls or API timeouts. The hiring manager wants to see pragmatic judgment—knowing when to aggressively hack together a prototype, when to harden the security boundary, and how you task multiple agents in parallel without letting technical debt spiral out of control.
Application Advice
Your resume needs to scream 'autonomous builder.' Do not list generic software engineering duties; highlight specific instances where you shipped production-grade AI agents, MCP integrations, or autonomous workflows from scratch as an individual contributor. Make sure to weave in keywords like Claude Agent SDK, MCP, RAG services, Python, and cross-functional stakeholder management directly into your work history to easily clear their ATS.