Forward Deployed Engineer

NimbleWay•San Francisco, CA

3d•$170,000 - $235,000•Hybrid

About The Position

This is Nimble’s founding Forward Deployed Engineer role. The engineer builds AI agents and search integrations with customers and takes them into production. Following qualification, the engineer partners with the Solution Consultant and customer engineers to refine the architecture and agree on success criteria. They build an initial working version using the customer’s real queries, workflows, and codebase, validate it against the customer’s evaluation set, lead implementation through production deployment, and hand off the work with the customer’s engineers equipped to operate and extend it. A typical engagement lasts 12 weeks. The role owns technical decisions for each engagement, including refining the implementation architecture, choosing an orchestration approach, defining evaluations, negotiating technical scope and sequencing, and determining what can be reused versus what should remain customer-specific. As the first member of the FDE team, the engineer also helps shape the function’s strategy, engagement model, playbook, and future hiring. The role is based in New York or San Francisco and requires working in the office three days a week. It reports to the Director of Solution Consulting and does not carry a sales quota. Base pay is $170,000 to $235,000 depending on level, with an additional 10% to 15% variable pay and equity.

Requirements

Write production Python and be able to read and modify TypeScript.
Have shipped and operated LLM-powered applications for real users and be able to explain how failures, latency, and cost were handled.
Have designed or improved evaluations for an LLM or another nondeterministic system and used results to make engineering decisions.
Have built tools for an agent to call using MCP, function calling, or an equivalent approach.
Have 4 or more years of software engineering experience.
Be able to show shipped LLM or agent work, such as a repository, live product, or case study.
Have worked directly with customers or external partners to scope and deliver technical projects, in a role such as FDE, software engineer, solutions engineer, consultant, or founder.
Be able to conduct a discovery call with a VP and debug a failing run with an engineer on the same day.
Be comfortable with ambiguity and rapid change in a fast-paced startup.

Nice To Haves

Experience with retrieval and RAG, including hybrid search, reranking, and freshness.
Experience building with LangGraph, the OpenAI Agents SDK, the Vercel AI SDK, or the Claude Agent SDK.
Experience delivering data to Snowflake or Databricks.
Experience with web data, such as scraping, anti-bot systems, and structured extraction.
Effective use of AI development tools, with responsibility for reviewing, testing, and maintaining the code they produce.

Responsibilities

Turn the customer’s goal into a working flow of Search, Extract, and Web Search Agent calls integrated into the customer’s code.
Build the first version using the customer’s real queries, working side by side with the customer’s engineer.
Define representative evaluation sets and success criteria with the customer; benchmark quality, latency, and cost during iteration; and instrument production workflows to detect failures and regressions.
Write orchestration code that keeps production flows stable, including fan-out, deduplication, retries, and cost caps.
Embed Nimble in the customer’s agent through its MCP server, SDKs, or API, on a schedule or on demand.
Present results to customer engineers and executives and record the technical win with the Solution Consultant.
Write the runbook, train the customer’s engineer, and hand the account to Customer Success.
Identify recurring customer needs and turn successful implementations into reusable integrations, evaluation tools, reference applications, and product improvements.
Shape the FDE function, including its engagement model, entry and exit criteria, playbook, and profile of future hires.