Skip to content

AI · Retrieval

RAG System

Private document Q&A system: ask questions across a document collection and get LLM answers grounded in the retrieved sources.

Backend With LLM · 2024 - 2025

Private codebase

What I built

  • Built a FastAPI pipeline covering document ingestion, preprocessing, metadata extraction, embeddings, vector storage, semantic retrieval, and LLM-based responses.
  • Stored document metadata in PostgreSQL and cached hot queries in Redis, with validation and background processing so large uploads never block the API.
  • Exposed search and question-answering APIs consumed by a React/TypeScript dashboard.
  • Containerised the services with Docker and deployed them on AWS.

Tech stack

PythonFastAPIPostgreSQLRedisPandasLLMs (Gemini, Llama)Vector DBReactTypeScriptDockerAWS

More projects

All projects

Full-Stack · Content Platform

Corporate CMS Platform

Brandcloud's corporate CMS, powering the company's live websites so content teams can publish pages, blogs, and media across sites without developer help - with LLM-assisted writing and SEO.

Read case study

Data · Automation

Competitors Ads Analysis AI

Competitive-intelligence tool that gathers competitors' ads into one place so marketing teams can review, classify, and report on them.

Read case study