Skip to content
Back to portfolio

Enterprise AI

Sage Enterprise AI Platform

A secure, RAG-powered enterprise AI platform that makes engineering knowledge discoverable across GitHub, Confluence, and Stack Overflow.

Senior AI Engineer at Eli Lilly and Company · January 2026 – Present

Public descriptions omit internal repositories, enterprise sources, credentials, prompts, endpoints, and proprietary data.

Problem

Engineering knowledge was fragmented across repositories and enterprise knowledge sources, making secure self-service discovery and onboarding difficult.

My contribution

Developed Python FastAPI backend services, implemented secure GitHub ingestion APIs, parsed Markdown content, generated embeddings for RAG, enforced RBAC, designed modular MCP integrations, implemented OAuth 2.0 with PKCE, integrated Stack Overflow and Confluence, built Microsoft Teams Bot Framework backend integrations with rich-card response support, supported continuous knowledge ingestion from engineering discussions, and contributed to CI/CD workflows. Currently generalizing the platform beyond a single engineering team into a multi-tenant product: designing self-service onboarding so any team can upload its own documentation and provision its own isolated, RBAC-scoped Sage instance.

Simplified architecture

Simplified: secure GitHub ingestion and Markdown parsing feed an embeddings-based RAG pipeline behind RBAC and OAuth 2.0 with PKCE, integrated with Stack Overflow, Confluence, and a Microsoft Teams bot for rich-card responses, with continuous ingestion from engineering discussions.

Challenges and tradeoffs

  • In progress: generalizing a single-team RAG assistant into a self-service, multi-tenant platform, where each team's documentation, embeddings, and Sage instance must stay isolated behind its own RBAC scope.

Technologies

PythonFastAPIRAGEmbeddingsRBACMCPOAuth 2.0PKCEGitHubStack OverflowConfluenceMicrosoft Teams Bot FrameworkMulti-tenant architecture