DocAsk — Internal RAG Policy Assistant by Akshay KathiriyaDocAsk — Internal RAG Policy Assistant by Akshay Kathiriya

DocAsk — Internal RAG Policy Assistant

Akshay Kathiriya

Completed work

AI Engineer

NestJS

Next.js

RAG

TypeScript

Chroma

Overview

DocAsk is a portfolio RAG (Retrieval-Augmented Generation) system that lets employees ask natural-language questions over company policy PDFs and receive answers with source citations.

Problem

Teams waste time searching long policy documents. Keyword search misses context; reading every PDF does not scale.

Solution

PDF ingest and chunking pipeline

Vector embeddings stored in Chroma

Semantic retrieval + LLM generation with citations

Next.js chat UI and NestJS API

MCP server for tool integration

Langfuse observability for tracing

Stack

NestJS · TypeScript · Next.js · Chroma · OpenAI / Claude / Ollama · Docker

Note

100% synthetic fictional company data — safe to show publicly. Live demo available on discovery calls via screen-share.

GitHub: https://github.com/KathiriyaAkshay/DocAsk

Like this project

Completed work

Posted Jun 14, 2026

Portfolio RAG system: ingest PDF policies, semantic search, LLM answers with citations. NestJS, Next.js, Chroma, multi-LLM support, MCP server. Synthetic data only.

Likes

Views

Timeline

Jan 1, 2025 - Jun 1, 2025