RESEARCH DEMO — NOT A PRODUCTION KYC SYSTEM
NNirikshaID Bench
About the work

A compact research-engineering system, end to end.

NirikshaID Bench connects dataset construction, model contracts, evaluation artifacts, failure analysis, an API, and a recruiter-friendly research interface.

Python core

Pydantic, Pillow, NumPy, RapidFuzz, FastAPI, and reproducible configuration.

Web system

Next.js, strict TypeScript, responsive CSS, accessible controls, and typed API data.

Deployment

Dockerized backend and frontend with environment-based API and CORS configuration.

Portrait of Ninad Naik
Author

Ninad Naik

CS Student | AI/ML Research and Engineering

Research interests: multimodal learning, document intelligence, computer vision, language models, and reproducible evaluation.

GitHubLinkedIn

Honest project status

The local deterministic baseline and full evaluation path are operational. Qwen inference and QLoRA are compute-dependent optional modules; no external-model score is shown unless that model is actually run.