Staff Software Engineer

JLL•Chicago, IL

23h•Remote

About The Position

JLL is rolling out AI agents directly into business lines: Capital Markets, Leasing, Property Management, and corporate functions. Forward Deployed AI Engineers (FDEs) are the engineers who sit inside those lines, learn the work from the people doing it, and bootstrap the agents that change how that work gets done. You will own a business line as a one-person engineering team. The Center of Excellence in our central AI team gives you the methodology, the platform, and the patterns; the business line gives you the problems, the SMEs, and the appetite for change. Your job is to turn that into shipped agents and MCP integrations, then hand mature work off to platform and operations as it stabilizes. We expect engineers at this level to operate with AI tools as a primary part of their workflow. Your leverage comes from the hooks, skills, workflows, and automations you install to ship safely to production. The leverage in this role is AI, not headcount.

Requirements

Shipped production agents using Claude Agent SDK, MCP, or an equivalent stack.
Production code in at least one of Python, Node, or .NET, with comfort moving across the others.
Worked with diverse data stores: relational, NoSQL, object/blob, vector, and document.
Operated end-to-end as a one-person team: design, ship, support, hand off.
Strong written and verbal communication; comfortable in the room with non-engineers.
Pragmatic judgment on when to ship, when to harden, and when to hand off.
Systems thinking with a focus on failure modes; AI writes most of the code, you own it.
Comfort with tasking and ability to run multiple agents in parallel as part of daily workflow.

Responsibilities

Build and ship agents for a specific business line, in close partnership with its SMEs and operators.
Stand up MCP servers and connectors: pre-production first, then through CDR into production.
Bootstrap skills, integrations, and evals on top of JLL's AI platform (LLM gateway, shared RAG services, MCP toolchain).
Run workshops with business teams, teach them to use Claude Code, and translate their workflows into agent skills.
Hand mature systems off to platform and operations teams once they outgrow the bootstrap phase.
Feed patterns, anti-patterns, and platform gaps back into the Center of Excellence so the next FDE moves faster.