Selected work

A few projects that show how I scope, build, and ship applied AI and workflow-heavy systems.

AI & Machine Learning(10)

AI/ML2024

Caption Art

AI-powered creative automation for visual content

Generative AI pipeline combining image analysis, style matching, and text generation via Fal AI.

ReactTypeScriptFastAPIFal AITailwind CSS
AI/ML2024

EchoAI-MLX

Fine-tuned Whisper on Apple Silicon

Fine-tuned Whisper on Apple Silicon using MLX. Benchmarked against base models for accuracy and speed on domain-specific audio.

PythonMLXWhisperApple Silicon
AI/ML2025

Advay Learning

MediaPipe-powered educational app for children with gesture recognition

Educational app with real-time hand tracking for children. Built with MediaPipe and Flutter for Android.

FlutterDartMediaPipePythonOpenCV
AI/ML2025

SceneGuide v3

Computer vision accessibility app for blind & elderly users (v3)

Accessibility app using CV to describe scenes and read text aloud for visually impaired users.

FlutterDartComputer VisionOCRTTS
AI/ML2025

model-lab

On-device audio AI — MLX, Whisper & model evaluation infrastructure

Production-ready evaluation framework for ASR/TTS models with automated scorecards and multi-device benchmarking.

PythonMLXWhisperApple SiliconCUDA
AI/ML2025

Waste Segregation App

AI-powered waste classification — Flutter app with Firebase & gamification

Published Flutter app with GPT-4.1-nano and Gemini Vision classification across 5 waste categories.

FlutterDartFirebaseOpenAIGemini Vision
AI/ML2025

Insurance RAG

RAG-based insurance document Q&A system

RAG pipeline with vector embeddings and LLM-powered Q&A for insurance policies with source citations.

PythonRAGLLMVector Database
AI/ML2025

Frame Analyser

Video frame-by-frame AI analyser — React, TypeScript, Vite

Video analysis tool with AI-powered frame inspection using React and TypeScript.

ReactTypeScriptViteAI
AI/ML2025

Agents Platform

Local-first multi-agent orchestration with human-gated risk tiers

Multi-agent platform with OpenAI, Anthropic, and Ollama runtimes. Human-gated risk tiers and local snapshot persistence.

TypeScriptNode.jsLangGraphOpenAIAnthropic
AI/ML2025

avia_new

Enterprise audio/video transcription platform with admin dashboard

Enterprise transcription platform with Whisper, spaCy NLP, and admin dashboard for business analytics.

PythonTypeScriptWhisperspaCySQLAlchemy