AI Software Engineer

We are looking for an AI Software Engineer who uses AI deeply in their own development workflow and can build dependable agents for the people running the business. This is a fully remote, full-time role (40 hours per week) with a required team overlap of 10 A.M. to 2 P.M. Eastern Time. You will build an agent that finds the right company facts, uses tools to solve an operational problem, explains its proposed action, and improves from reviewed corrections.

Key Responsibilities

Agent Runtimes & Tools: Build agents with persistent sessions, structured tool calls, browser or code execution, sandboxing, scoped permissions, retries and clear approval boundaries.

Company Knowledge & Memory: Connect documents, APIs and databases. Design retrieval with source citations, freshness and access controls, and keep verified facts, conversation state and learned preferences separate.

Measurable Improvement Loop: Use traces, test cases and user corrections to improve prompts, tools, retrieval and memory. Test changes before promotion and keep a way to roll them back.

End-to-End Products: Ship chat and workflow interfaces tied to real systems, from guest support and company knowledge to scheduling and operations. Own testing, deployment and debugging.

Quality in Production: Create evaluations, inspect raw traces, reproduce edge cases and measure task success, latency and cost. Diagnose failed tool calls or retrieval results and explain the fix.

Solid Software Foundations: Build APIs and frontends on strong engineering basics, including concurrency, queues, idempotency, tests, Docker and CI/CD.

Qualifications & Requirements

Coding Agents: Hands-on use of tools such as Codex, Claude Code or OpenCode beyond autocomplete, and the ability to explain how you manage context, parallel work, verification and failures.

Agent Stack Knowledge: Deliberate choice of models and orchestration. Experience with MCP, APIs or code-mode tools, and an understanding of context limits, state persistence, tool schemas and when deterministic code is the better choice.

Programming: Strong Python and/or TypeScript, plus SQL and document databases.

Proof of Work: A resume or profile, GitHub or a demo, and one agent system you personally built: what you owned, how it works, one failure you fixed and how you measured improvement.

Schedule: Fully remote, full-time at 40 hours per week, with team overlap from 10 A.M. to 2 P.M. Eastern Time.