Pre‑GPT self‑taught engineer.
Building things I’d want to use. Interested in products, systems & hard problems
Currently exploring OpenClaw and AI Agents and building a prompt-to-animation video generator and a generative game engine.
I’m always open to interesting project collaborations and remote opportunities.
Python · JS/TS · Parlant · PageIndex · React · Svelte · MongoDB
Jalandhar, India (Hybrid) - Jan 2026 – Present
Identified 22 security vulnerabilities (1 critical) across 50+ APIs and 523 pages, impacting 60k+ MAU. Built and deployed a custom Shopify web application, improving UI/UX and resolving SEO/GEO performance issues. Built VLM-based KYC pipeline, developed and maintained Next.js web apps and Python scripts for automation. Developed the personal website for the CEO.
Delhi, India (On-site) - May 2025 – Aug 2025
Engineered data pipelines processing 100k+ records, automating product listing, reporting, and data cleaning workflows. Automated address generation workflows, saving 1 hour/day per employee and reducing operational overhead. Built scalable scraping and automation systems to process large CSV/Excel datasets.
Punjab, India - B.Tech CSE, Minor in DevOps - 2023 – 2027
Full Stack Development, DevOps & Web3 - 2024 – 2025
Infinite-canvas chat for non-linear LLM workflows with branching and structured exploration, now extended with vectorless RAG, subagent orchestration, and tool calling.
Multi-strategy NIFTY 50 options trading system with concurrent live execution and centralized data pipelines. Architected modular alpha, execution, and risk layers with dynamic SL/TSL mechanisms. Achieved ∼60% directional accuracy in live trading using multi-factor signal models and adaptive risk control.
Real-time handwriting recognition system using custom CNNs trained on MNIST and HASYv2 datasets. Designed a Union-Find–based stroke grouping system for accurate multi-stroke mathematical symbol recognition. Delivered a high-performance, responsive UI optimized for real-time interaction.
Full-stack multi-vendor marketplace with modular services and role-based dashboards. Implemented RESTful APIs with JWT-based authentication, refresh-token rotation, and RBAC authorization. Hardened backend using Helmet, rate limiting, and input validation to improve security and reliability in production.
Implemented a spring physics-based motion easing library and visualizer for smooth UI animations. Published as motionspring on NPM.
Neural network implementation from scratch with forward pass, backpropagation, and training loop.
Utility for batching LLM calls to reduce cost and improve request efficiency.
Browser extension bridging ChatGPT web sessions to local development environments, enabling agentic coding workflows directly from the browser window.
API for generating transparent PNG images using AI system prompts and chroma keying.
Camera2API camera app with full manual control, bypassing OEM post-processing for ungraded output. Implemented custom log gamma + Bézier tone curves for dynamic range preservation and post-processing flexibility.
CLI that retrieves live web content and generates concise summaries using Gemini AI. Implemented a data extraction pipeline with Trafilatura to clean, normalize, and render structured markdown output.
Multithreaded CLI tool for generating videos from spec-based scripts with layered audio tracks and SFX.
Lightweight image-processing CLI for adjusting brightness, saturation, and contrast.
2026-03-28
It has never been easier to build a product. Coding is no longer the bottleneck. The bottleneck is figuring out what people actually want (w...
2026-03-10
Code used to be the hard part of software (was it?). Now it isn't. The hard part is building systems that don't break (or rather break grace...
2026-02-15
When prompting AI models, token count directly affects both cost and latency. This benchmark compares how efficiently different languages ex...
2026-01-22
Started coding around 2014–15; I don't remember exactly. Why? Games! My first language was MSW Logo, then QBASIC also VBA. Then C++; here’s...