James Shehan
About
I'm an AI product strategist turned builder — 8+ years driving digital transformation in financial services and enterprise AI, now shipping production systems myself. At Right at Home, I architected and launched enterprise AI products — Rahya (NL-to-SQL analytics) and CIA (corporate knowledge agent) — across a 700+ location franchise network while leading AI strategy, governance, and enablement programs. I've also built 3 personal AI projects from scratch: a real-time voice coaching agent, an AI wine concierge with two-tier RAG, and a self-hosted multi-agent infrastructure node — 5 AI products shipped and counting. I hold a PMP, Azure AI Engineer certification, and a Certified Sommelier credential — because the best technologists pursue mastery across domains.
Work Experience
Education
Skills
Check out my latest work
Personal AI projects and enterprise products I've built end-to-end — from architecture and system design through production deployment.
Pitch Coach AI
Built a real-time AI sales coaching voice agent with a two-brain architecture: a real-time actor handles live conversation via Azure OpenAI gpt-realtime, while an async grader evaluates performance against rubric-based criteria. Features dynamic JSON buyer personas, structured feedback, and LiveKit for low-latency voice streaming.
Vinny — AI Wine Concierge
Designed and built a full-stack AI wine recommendation engine with two-tier RAG across 71K wine reviews and 181 Wikipedia articles. Features MCP server integration, 10 specialist agents, Supabase with pgvector for semantic search, rate limiting via Upstash Redis, and enterprise-grade CSP security headers.
Hive — Multi-Agent Infrastructure
Architected a self-hosted multi-agent infrastructure node with Docker sandboxing, 6-layer security model, and 19 Architecture Decision Records. Features LiteLLM cost governance, QMD hybrid search, Tailscale zero-trust mesh networking, and cloud-init with LUKS + TPM2 encryption for bare-metal deployment.
Rahya — Enterprise Data Intelligence
Led product strategy and rollout of an enterprise NL-to-SQL analytics platform that democratized data access for non-technical users across a 700+ location franchise network. Built on Snowflake Cortex with a natural language interface that translates business questions into verified SQL queries, projected to reduce annual data analysis costs by $150K+. Managed full lifecycle from pilot kickoff through corporate launch including training, onboarding, and adoption tracking.
CIA — Corporate Knowledge Agent
Launched an enterprise knowledge retrieval agent deployed via Microsoft 365 Copilot, serving as the central 'front door' for corporate information access. Built with RAG architecture on Azure for secure, grounded responses from internal SharePoint knowledge bases. Operationalized pilot testing, iterative feedback loops, and adoption campaigns to scale usage across corporate employees.