Hakeem Shitta-Bey

Created a document preprocessing pipeline to improve downstream AI understanding, combining parsing, normalization, and tool/model-based extraction for structured representations. Generated a hypergraph from SRL outputs and transformed it into an event graph using heuristic methods and BERT-family models (incl. LegalBERT); trained GNNs to produce embeddings for a Pinecone-backed RAG system to improve grounding and reduce failure modes. Scoped the engagement independently with tasks definition and end-to-end execution with minimal oversight.

Implemented a notification workflow in a Temporal-based service, configuring provider logic and workflow steps for customizable in-app delivery. Built gRPC service methods to communicate with a Postgres-backed internal API, improving integration between backend services and platform tooling. Contributed to a React/TypeScript frontend by cleaning up legacy code, fixing UI bugs, and shipping minor improvements for feature parity.

Led end-to-end development of a full-stack mobile app for student resources, events, social feeds, and collaboration using React Native (Expo), TypeScript, Go, and PocketBase (migrating to Supabase). Implemented a tool-calling chatbot backed by a vector database and a RAG pipeline to retrieve personalized context and execute actions via backend APIs. Engineered a scalable data model with 15+ collections and built a fast, memoized UI with virtualized lists; implemented real-time messaging via SSE (migrating to WebSockets). Managed product, engineering, and operations for a 10+ person team and drove campus rollout to 200+ users through partnerships.