Fractional AI Strategy Advisor + Infrastructure Consultant (Python/LangGraph/Docker)

Role Overview

We are seeking a senior AI strategy advisor for a solo financial advisory practice. This is a fractional, long-term advisory engagement — not a project-based contract. You will serve as a trusted technical advisor and occasional hands-on implementer, working directly with the founder/architect.

The practice runs a local Docker-based AI infrastructure (13 containers, 5 profiles) built on LangGraph workflows, PydanticAI agents, LiteLLM multi-model routing, MCP (Model Context Protocol) bridges, Graphiti knowledge graph (FalkorDB), and RAGFlow document search. The system operates under FINRA compliance requirements with fail-closed PII redaction.

What You Will Do

- Advise on architecture decisions, tool selection, and AI/LLM best practices as the landscape evolves
- Guide implementation of new capabilities — you know how to do things that would take the founder hours of research
- Review and improve existing LangGraph workflows and PydanticAI agent designs
- Recommend and evaluate new frameworks, models, and orchestration patterns
- Occasionally implement features hands-on when advisory alone is not enough
- Help with operations and maintenance — upgrades, monitoring, infrastructure care

Required Skills

- Deep expertise in the AI/LLM ecosystem — frameworks, models, orchestration patterns
- Strong Python skills (FastAPI, async, LangGraph SDK, PydanticAI)
- Docker Compose and self-hosted infrastructure experience (NOT cloud-only)
- Ability to track the fast-moving AI landscape and translate trends into actionable guidance
- Excellent communication — you can explain complex technical tradeoffs clearly

Strong Plus

- LangGraph production experience
- MCP (Model Context Protocol) — building or consuming MCP servers
- Knowledge graphs (Graphiti, FalkorDB, Neo4j)
- RAG pipeline design and optimization
- Multi-model routing (LiteLLM or similar)
- Experience advising solo practitioners or small teams (vs. enterprise)

Engagement Model

- Fractional, ongoing — approximately 10-15 hours/week
- Month-to-month, flexible working style (async, sync, pair sessions)

- You are the expert advisor; the founder is the architect and decision-maker