Senior AI Builder / AI Engineer

The Role

We're looking for a Senior AI Builder, someone who doesn't just understand AI systems in theory but ships them in practice. You'll design, build, and operate production-grade AI agents and intelligent workflows that solve real business problems.

This role sits at the intersection of software engineering and applied AI: you'll wire up autonomous agent pipelines, integrate retrieval-augmented generation systems, orchestrate complex multi-step workflows, and push the boundaries of what's possible with the latest Claude toolchain.

The ideal candidate has deep hands-on experience with the Anthropic ecosystem (Claude API, Claude Code, and Managed Agents) and pairs it with strong Python fundamentals, database fluency, and an instinct for reliable, production-ready systems.

Required Skills & Expertise

Claude Code & Managed Agents Production experience building and deploying autonomous agents using Anthropic's Managed Agents API, including session management, multi-agent coordination, sandboxed execution, and custom tool definitions. Comfortable using Claude Code as a primary development environment for agentic coding workflows.

Familiarity with Claude Cowork (Anthropic's agentic desktop environment) as both a user and a builder:

* Creating custom Skills and Plugins

* Configuring MCP connectors and scheduled tasks

* Leveraging agentic capabilities to automate workflows across business functions

Python Development 3+ years shipping Python in production. Strong command of:

* Async patterns and type hinting

* Modern tooling: uv, ruff, pytest

* Building APIs with FastAPI or Flask

* Writing clean, testable, well-documented code

RAG Systems Working experience implementing retrieval-augmented generation pipelines, including:

* Embedding models and vector databases (pgvector, Pinecone, or similar)

* Chunking strategies and hybrid keyword + semantic search

* Evaluating and improving retrieval quality, not just wiring up the plumbing

Supabase & PostgreSQL Proficient with Supabase as a backend platform:

* PostgreSQL layer, Row Level Security, Auth, Edge Functions, Realtime

* Writing complex SQL, managing migrations, optimizing query performance

What You'll Do

* Deploy and manage Claude Managed Agents for long-running, autonomous workloads, including environment configuration, tool scoping, and session monitoring

* Architect and ship end-to-end AI agent systems from initial prompt engineering through production deployment and monitoring

* Build and maintain RAG pipelines that deliver accurate, grounded responses over proprietary data sources

* Develop and extend Supabase-backed data layers: migrations, RLS policies, and performant queries that scale

* Build custom Claude Cowork Skills and Plugins that bring AI capabilities to non-technical teams across the organization

* Collaborate with product, design, and business stakeholders to translate ambiguous requirements into working AI-powered solutions

* Establish best practices for prompt engineering, agent evaluation, guardrails, and responsible AI deployment

Nice to Have

* n8n: building automated workflows, connecting APIs, conditional branching, error recovery, and webhook integrations

* Message queuing: RabbitMQ or Redis (Streams / Pub-Sub) for event-driven architectures and reliable task coordination

* MCP: Model Context Protocol server development and integration

* Multi-model orchestration: familiarity with additional LLM providers and routing strategies

* Containers: Docker or Kubernetes for agent deployment

* TypeScript / JavaScript : for Supabase Edge Functions or lightweight tooling

* Observability: OpenTelemetry, Datadog, or Grafana for monitoring agent performance

* Open-source AI contributions or published work in applied AI

Tech Stack

Python · Claude API · Claude Code · Managed Agents · Claude Cowork · RAG / pgvector · Supabase · PostgreSQL · n8n · MCP · FastAPI

ps. If you're the kind of engineer who's already building agents in your spare time, who reads API changelogs for fun, and who thinks "autonomous" is an engineering requirement rather than a buzzword, we want to hear from you.