At Topmate, I developed a semantic recommendation service that matched user profiles with relevant mentors and platform resources. I built FastAPI endpoints for profile processing, embedding generation, candidate retrieval, and ranked recommendations.
I also implemented preprocessing and similarity-ranking pipelines for heterogeneous user and content data, and integrated model inference with backend services.
On Splitty, I built a multimodal receipt-scanning feature with the Gemini API that generates structured JSON expense entries. I also engineered a debt-consolidation algorithm and designed PostgreSQL schemas using Prisma ORM.
For my Enterprise RAG API & Semantic Search Engine project, I developed a backend for document parsing, context indexing, and semantic Q&A with ChromaDB and HuggingFace embeddings. I also built asynchronous API endpoints for cloud LLM inference and containerized the deployment pipeline with Docker.
