SHU.ai
Principal Software Engineering Manager
2026
As a founding member of SHU.ai — an early-stage private AI infrastructure startup — I helped build the product from the ground up: frontend, backend, AI/RAG workflows, customer onboarding, billing, and production infrastructure. A look at what I built.
Public web presence
Owned significant portions of shu.ai — implementing pricing and signup integrations, executing major site redesigns, preparing production releases, and adding search-engine optimization.
Production chat experience
Built the core chat surface — configurable assistant branding, model selection, contextual onboarding, streaming with cancellation, and durable handling of disconnected clients.
Onboarding & billing
Implemented customer onboarding and signup flows with billing integration, moving new users from landing page to a working private AI workspace.
Personal Knowledge Base
Owned the architecture and 0-to-1 delivery of the Personal Knowledge Base — letting users create private knowledge, attach it directly to AI conversations, persist it across sessions, prevent duplicate ingestion, and turn conversations into reusable knowledge.
Knowledge in context
Made a user's private knowledge visible and controllable inside the conversation — persistent attachments and session-context visibility that ground answers in their own documents.
Under the hood
Beyond the surfaces above, I owned foundational platform work:
AI usage & cost observability
Architected customer and administrator experiences for tracking credits, tokens, requests, model usage, and plan consumption — with per-user/model cost drilldowns and provider-sourced LLM pricing.
Resilient RAG & streaming
Improved scalability and reliability across core AI/RAG workflows by eliminating long-lived database transactions around LLM streaming, embeddings, retrieval, Stripe, and external calls; strengthened chat-stream lifecycle handling, provider fallback, and production failure recovery.