Software / Forward Deployed Engineer

Client is embedding AI agents directly into Legal systems and the infrastructure that runs them.

You'll own the Legal business line as a one-person engineering team. The Center of Excellence, part of the central AI organization, provides the methodology, platform, and proven patterns; Legal provides the problems, the subject matter experts, and the appetite for change. Your job is to turn that combination into shipped agents and MCP integrations, then hand mature work off to platform and operations teams as it stabilizes.

We expect engineers at this level to treat AI tools as a core part of their workflow, not an occasional aid. Your leverage comes from the hooks, skills, workflows, and automations you build to ship safely into production.

What You'll Do

• Build and ship agents in close partnership with subject matter experts and operators.
• Stand up MCP servers and connectors, moving from pre-production through design review into production.
• Bootstrap skills, integrations, and evaluations on top of the AI platform, including the LLM gateway, shared RAG services, and MCP toolchain.
• Run workshops with teams, teach them to use Claude Code, and translate their workflows into agent skills.
• Hand off mature systems 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 engineer moves faster.

Must-Have Experience

• Shipped production agents using Claude Agent SDK, MCP, or an equivalent stack.
• Production experience in at least one of Python, Node, or .NET, with the ability to move across the others.
• Worked with a range of data stores, including relational, NoSQL, object/blob, vector, and document databases.
• Operated end-to-end as a one-person team: designing, shipping, supporting, and handing off work.

Traits

• Strong written and verbal communication, with comfort presenting to 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, but you own the outcome.
• Comfort tasking and running multiple agents in parallel as part of your daily workflow.