AI Software Engineer

Our client is hiring engineers who turn large language models into dependable software that real users rely on every day. You’ll own AI features end to end: understand the workflow, design the agent or retrieval system, ship it, measure it, and keep it healthy in production.

This is a software engineering role first. The hard problems aren’t “can the model generate an answer?” but control, grounding, reliability, latency and cost. You’ll decide where an LLM adds value, where plain code is safer, and where a human should stay in the loop.

What you’ll do

• Design, build and ship LLM-powered features: customer-facing products and internal tools that automate real workflows
• Architect agentic systems: tool calling, structured outputs, multi-step orchestration, memory and context management, human-in-the-loop approvals, MCP servers
• Build retrieval pipelines (RAG): ingestion, chunking, embeddings, hybrid search, reranking, with strict data isolation
• Make quality measurable: golden datasets, regression suites, LLM-as-a-judge, release gates; turn production failures into eval cases
• Run it in production: backend services and APIs (Python and/or TypeScript) on AWS, Azure or GCP, with retries, fallbacks, guardrails, tracing and cost optimization
• Use AI coding agents (Claude Code, Cursor, Copilot) daily, and critically review their output
• Work directly with product and business teams to find high-value problems and take them from idea to adoption

• 5+ years of professional software engineering with recent ownership of production code (strong 3-4 year profiles with exceptional AI delivery are welcome)
• Strong Python and/or TypeScript/Node.js
• Shipped LLM-powered systems that stayed in production after launch: agents, RAG, copilots or automations. Not prototypes or weekend projects
• Hands-on agentic engineering: tool calling, structured outputs, prompt and context design, orchestration frameworks (LangGraph, PydanticAI, Semantic Kernel or similar)
• Practical RAG experience: embeddings, vector search (pgvector, Qdrant, Pinecone, OpenSearch), chunking, reranking, and knowing where each breaks
• Evaluation discipline: you’ve built eval sets and monitoring, and can explain how you knew an improvement was real
• Solid backend fundamentals: REST APIs, PostgreSQL, Docker, CI/CD, and at least one managed LLM platform (OpenAI, Anthropic, Bedrock or Vertex)
• Clear professional English and the ability to explain trade-offs to non-technical people
• Reliable remote setup with meaningful overlap with US working hours

Nice to have

Nobody has all of these; depth in one or two makes you stand out. Fine-tuning or serving open models, LLM observability (Langfuse, LangSmith), data engineering for unstructured data, voice or multimodal pipelines, Kubernetes/Terraform, React/Next.js for AI interfaces, regulated domains (fintech, healthcare, legal), or founding-engineer experience.

Who you are

• A builder who goes from vague problem to working solution in days, then iterates on real usage
• An owner who ships it and stays with it after launch
• A systems thinker who cares how components interact, fail and scale, not just about prompts

This probably isn’t for you if

• Your AI experience is mainly tutorials, hackathons or a thin wrapper around a chat API
• You prefer pure research with no production responsibility
• You need a detailed spec before you can make progress
• You’d rather tweak prompts than find the real root cause

• $200,000 - $250,000 base salary (W-2), depending on experience
• Fully remote, flexible hours
• Real ownership on small, senior teams
• Access to frontier models and AI coding tools

We welcome applicants of every background., * Dental insurance

• Health insurance
• Paid time off
• Vision insurance