JavaScript is my first language and still my favourite. React on the frontend, Node + Express on the backend, SQL when data needs structure. That's my home base.

Outside of that — I reach for whatever the project needs. Built ML-powered health tools using Python + scikit-learn (with a lot of help from docs and AI tooling, I'll be honest). Integrated ViT models for deepfake detection. Wired Solidity contracts to frontends via ethers.js. I'm not an ML researcher or a Solidity expert — I'm a JS developer who isn't scared to go outside his comfort zone when the problem is interesting enough.

Right now I'm building ComplianceOS — a multi-agent AI platform for India's MSME sector — and learning blockchain internals properly, not just copy-pasting contract code.

Delhi NCR · Top 10 @ CodeWizards (100+ teams) · Open to collabs and freelance

ComplianceOS  ·  React Node.js Express PostgreSQL MongoDB Groq LLM Redis

India has 6.3 crore MSMEs drowning in compliance paperwork. I built five AI agents — ARIA for onboarding, VEDA for legal lookup, SCOUT for scheme discovery, PATHWAY for portal navigation, SENTINEL for deadline tracking — all orchestrated together. Not a chatbot with a FAQ. An actual multi-agent system.

DeepTrust AI  ·  code  ·  React Flask ViT (HuggingFace) Solidity ethers.js Sepolia

"Just trust the AI" felt like a bad answer for deepfake detection. So I added a second layer — every verdict gets hashed on-chain via MetaMask. Upload face → AI analyses → result recorded on Ethereum Sepolia. Immutable proof, not just a confidence score. Integrated the ViT model through HuggingFace's API; the Web3 wiring is mine.

Healthcare Ecosystem  ·  Top 10 · CodeWizards  ·  React FastAPI XGBoost Solidity Hardhat ethers.js

Built under 48h for a hackathon. Fall-risk prediction model (XGBoost, pre-trained — I wired it into the app and visualised the SHAP outputs), patient consent on-chain, five AI agents for triage and resource flow, Telegram alerts for critical events, eleven role-based portals. Top 10 out of 100+ teams.

AI HealthMate  ·  code  ·  React FastAPI Firebase Scikit-learn Recharts

Diabetes risk prediction + mental health assessment in one place. Used XGBoost for prediction, built the full React frontend and dashboard myself. The ML part runs via an API — I know how to use the models, build around them, and make them useful. Not claiming to tune hyperparameters for fun.

LevelUp Journal  ·  code  ·  Flask SQLite SQLAlchemy Vanilla JS

Built this for myself — mood tracking, code progress logs, reflection prompts, particle background. Wanted something that made daily logging feel less like a chore. I use it. It works.

Weather Vibes  ·  code  ·  HTML CSS JS Tomorrow.io OpenCage

Weekend project. Real-time weather, live local time, animated day/night transitions. Started small, went a bit overboard with the animations. No regrets.

// home base
const stack = {
  languages : ["JavaScript", "Python", "C/C++"],
  frontend  : ["React", "TailwindCSS", "Framer Motion", "HTML/CSS"],
  backend   : ["Node.js", "Express", "Flask", "FastAPI"],
  databases : ["PostgreSQL", "MySQL", "SQLite", "MongoDB"],
  tools     : ["Git", "Prisma", "Firebase", "Redis", "REST APIs"],
};

// comfortable with, used in real projects
const alsoKnow = ["XGBoost", "scikit-learn", "HuggingFace APIs", "Recharts", "ethers.js"];

// actively learning
const learning = ["Solidity", "smart contract internals", "system design", "DSA"];

   

building something interesting? let's talk.
richhariyadivyank1@gmail.com