Skip to content

enterprise-rag-reliability-platform.md

Latest commit

 

History

History
58 lines (44 loc) · 2.43 KB

enterprise-rag-reliability-platform.md

File metadata and controls

58 lines (44 loc) · 2.43 KB
Raw

Enterprise RAG Reliability Platform

This is the flagship applied AI story in my portfolio.

The implementation lives primarily in ai-reliability-lab, with the public live demo surface supported by applied-ai-eval-lab.

I use "enterprise RAG" here to describe the workflow shape, not external adoption. The project is a portfolio-scale system built around the reliability pieces I would expect before treating a document AI workflow as more than a demo:

  • document ingestion and chunk metadata
  • retrieval with visible source evidence
  • cited answers
  • refusal behavior when evidence is missing or sensitive
  • evaluation cases and release-gate style checks
  • traces, metrics, and report artifacts
  • dashboard and CLI flows
  • deterministic no-key verification before optional model providers

How The Pieces Fit

Surface Role
ai-reliability-lab Full backend/platform repo: FastAPI, SQLite, retrieval, providers, evals, traces, dashboard, CLI, proof artifacts
applied-ai-eval-lab Live document intelligence demo: static dashboard, cited answers, answer-fact coverage, release-gate checks, report artifacts
Portfolio proof ledger Public verification map for local tests, demo checks, and Docker-free proof paths

Why It Matters

RAG demos can look convincing while hiding the important questions:

  • Did retrieval find the right evidence?
  • Does the answer cite the evidence it used?
  • Does the system refuse unsupported questions?
  • Can evaluation catch regressions before a provider or prompt change?
  • Can a reviewer run the core workflow without API keys?

The goal of this project is to make those questions inspectable.

Review Path

  1. Start with the full platform repo: https://github.com/GoparapukethaN/ai-reliability-lab
  2. Open the dashboard walkthrough: https://github.com/GoparapukethaN/ai-reliability-lab/blob/main/docs/demo.md#dashboard-demo
  3. Try the live Applied AI Eval Lab demo: https://goparapukethan.github.io/applied-ai-eval-lab/
  4. Check the proof ledger: https://github.com/GoparapukethaN/kethan-portfolio/blob/main/docs/proof-ledger.md

Honest Boundary

This is not a claim of production users, customer adoption, or business impact. The signal is the engineering workflow: code, docs, tests, evals, dashboard behavior, artifacts, and tradeoffs that I can explain in detail.