DS
PORTFOLIO SHOWCASE

Crafted projects.

Click any project card to open its interactive 3D Glass Folder modal, deep system specs, and live code repositories.

AI & Audio ML / Cybersecurity

SoundProof

AI Deepfake Audio Detector

SoundProof detects whether any audio file contains a real human voice or AI-generated synthetic speech in under one second, with a confidence score.

PythonPyTorchLibrosaStreamlitASVspoof 2019CNN
Fintech & Legal NLP

ClauseLens

Generative AI Legal Contract Risk & Citation Alignment Engine

Manual contract reviews are time-consuming and prone to human oversight. ClauseLens solves this by combining Generative AI with a verifiable citation alignment engine.

React 19TypeScriptViteTailwind CSSGeminiGroq LLMsPostgreSQL2FA OTP
Computer Vision / AI Forensics

DiffGuard

Production Reimplementation of IEEE/CVF WACV 2025 DiffFake

DiffGuard is a production-grade deepfake detection platform implementing the DiffFake architecture (arXiv:2502.16247). Instead of framing detection as a binary classification problem (which overfits to known fake generators), DiffGuard extracts deep facial embeddings from time-separated frame pairs, fuses them, and models the natural distribution of real-pair variation using a Gaussian Mixture Model (GMM).

PyTorchEfficientNet-b4Gaussian Mixture ModelsGrad-CAMNext.js 14FastAPICeleryRedisDockerPostgreSQL