Full-Stack Developer · ML Engineer · Building things that matter
ria = {
"currently_building": ["Velaris 💎","TryOn"],
"research_projects": ["WYA", "Glowing Guide", "TryOn"],
"shipped": ["WYA 🎨", "Luna ✨"],
"studying": "CS & Data Science @ MUJ (3rd year)",
"interests": ["Full-Stack Dev", "Machine Learning", "Data Science"],
"stack": ["Python", "React", "FastAPI", "Scikit-learn"],
"fun_fact": "I build apps to solve problems I actually have 🚀"
}Ever felt like your style exists but you just can't name it?
WYA is a full-stack AI-powered wardrobe intelligence platform. It analyses your wardrobe, builds a Style DNA profile, generates outfit combinations, and surfaces wardrobe gaps — all backed by a Deploying WYA — full-stack AI wardrobe platform on AWS
✦ FastAPI backend on AWS EC2 ✦ FashionCLIP AI tagging pipeline
✦ JWT authentication ✦ Style DNA engine
✦ Gap analysis with affiliate mapping ✦ Outfit generation
✦ Sustainability scoring ✦ CloudFront + S3 deployment
Talk to your wardrobe in plain English.
Luna is a standalone conversational frontend that connects to WYA's fashion intelligence API. Instead of navigating menus, you just talk — Luna classifies your intent, routes it to the right WYA endpoint, and responds with real outfits, real wardrobe gaps, and real style insights from your actual closet.
✦ Natural language intent classifier ✦ uses LLM to answer any fashion related question
✦ Routes to WYA's real API ✦ React + TypeScript
✦ Wardrobe search, outfit help ✦ Deployed on AWS S3 + CloudFront
✦ Gap analysis, Style DNA chat ✦ JWT auth via WYA
Type it. Sketch it. Upload it. Get a jewellery design you can actually build.
Velaris is an AI-powered jewellery design tool that transforms natural language descriptions, sketches, or inspiration images into professional design concepts and manufacturer-ready briefs — bridging the gap between a customer's idea and a jeweller's workflow.
✦ Text / sketch / photo input ✦ AI concept image generation
✦ Structured specification sheets ✦ Exportable PDF design briefs
✦ Manufacturer-ready output ✦ Planned: multi-view & cost estimation
Core flow: User input (text/sketch/photo) → AI concept image → structured spec sheet (type, metal, stone, cut, style, setting, occasion) → exportable PDF for jeweller quotation.
| Project | Description | Tech |
|---|---|---|
| ✍️ flowWrite | Multi-pass NLP pipeline that rewrites AI-generated text into natural human-like writing with tone control & HLS scoring | Python · NLP |
| 📚 BookNest+ | Smart reading platform with mood-based & behavior-driven book recommendations | TypeScript · React · ML |
| ₿ Bitcoin Fraud Detection | Graph Neural Network model that detects fraudulent bitcoin transactions using temporal & relational patterns | Python · GNN · PyTorch |
| 📱 Smart Resale | Predicts fair resale prices for used phones & laptops | Python · Pandas · Scikit-learn |
Languages
Frameworks & Libraries
Cloud & DevOps
Tools
- 🎨 Deploying WYA — full-stack AI wardrobe platform on AWS
- ✨ Building Luna — conversational AI stylist powered by WYA's API
- 💎 Building Velaris — AI jewellery design tool, idea to manufacturable brief
- 🎯 Deepening ML knowledge: GNNs, recommendation systems, embeddings
- 🌐 Exploring backend architecture, API design & cloud infrastructure