← Selected WorksGitHub

AI / LLM / RAG · 2025

Production RAG Chatbot

Production RAG chatbot for IAT Networks — 0% injection bypass.

PROD
// Overview

A fully productionised RAG chatbot deployed for IAT Networks, a regional ISP. The retrieval pipeline chains query expansion through MiniLM embeddings into ChromaDB (top-8 retrieval), followed by cross-encoder reranking to surface the top-4 most relevant chunks before handing off to Groq's SSE streaming API. A 4-layer GuardRail stack — injection regex, PII detection, domain filter, and toxicity gate — ensures no adversarial prompt escapes. Dockerised on Railway with zero downtime across all deployments.

// Specs

Specifications

ClientIAT Networks (ISP)
TTFT<800ms time-to-first-token
Security0% injection bypass
UptimeZero downtime (Railway)
RetrievalChromaDB top-8 → cross-encoder rerank top-4
// Features

Features

01

Query expansion → MiniLM embedding → ChromaDB top-8 retrieval pipeline

02

Cross-encoder reranking to top-4 for precision context delivery

03

4-layer GuardRail: injection regex, PII filter, domain filter, toxicity gate

04

Groq SSE streaming responses — <800ms time-to-first-token

05

Dockerised on Railway with zero-downtime continuous deployment

06

0% injection bypass rate across full adversarial test suite

// Tech

Tech Stack

ChromaDBGroqFastAPIDocker

Interested in
Production RAG Chatbot?

Explore the source code, architecture notes, and full implementation.

View on GitHubGet in Touch →