Work · Index
20 projects and 16 lab experiments. Each one has a case study with the stack, my role and the numbers I can actually back up.
My mother tongue had no search engine for its greatest poet, and Roman Sindhi has no standard spelling. Could retrieval still find the right verse?
A retrieval-augmented search engine over Shah Jo Risalo, the 18th-century Sindhi Sufi poetry of Shah Abdul Latif Bhittai.
2,519
verse documents indexed
FAISS IndexFlatL2, 384-d multilingual embeddings
Buyers and sellers of physical commodities needed live prices, matched orders and payments in one place, not spreadsheets and email.
A real-time global commodity trading platform with a matching engine, carbon-credit listings, live WebSocket data and automated trading bots.
38+
REST endpoints
Plus 8 auth routes and a dedicated WebSocket channel
Rate limiting should not cost more latency than the request it protects.
A framework-agnostic rate-limiting library for Node.js and Bun: 8 framework adapters, 8 storage backends, millions of checks per second.
5.5M+
ops/sec on Bun
4M+ on Node.js, 10K-key benchmark
Searching your own gallery shouldn’t mean uploading it to someone else’s server.
An iOS-style photo gallery for Android that understands your photos with on-device AI, without sending anything anywhere.
0
network permissions
No Internet access, accounts, ads or analytics
A Maps list tells you a business exists. It doesn’t tell you what it’s struggling with, or who to call about it.
Type “Cafes in Dayton, Ohio” and get sales intelligence back: every business found, enriched, audited and prioritized, with a client-ready audit PDF for each.
8
pipeline stages
Discover to export, checkpointed item by item
Most scrapers are one-off scripts. Making one safe to run for strangers, on any URL they type, is a different problem.
A generic web scraper SaaS: point it at a URL, it detects listings and pagination, you map fields, and it exports clean data. Hardened for multi-tenant use.
~4×
faster detail-page fetching
Bounded concurrency instead of a sequential loop
The lab
Personal tools, experiments and code generators. Nobody asked for these. That’s why they’re fun.
A two-stage OCR + LLM pipeline that pulls structured JSON out of receipts and invoices with no training at all.
A private app that turns personal life logs into a living graph of the people you know and how much you trust them.
I have rebuilt my portfolio six times. This site is version six, and one version taught itself which phrases to highlight.
A lofi music generation and post-processing dashboard: generate, blend, polish and export tracks from one place.
A donation platform that connects donors with people who need help, through verified accounts and reviews.
Cleans up messy workspace folders: a token-dense map an LLM can audit, and a Python script that applies the layout it proposes.
A browser extension that marks, row by row, which map leads have an email or phone number, so you qualify them at a glance.
A notebook pipeline that detects, encodes and matches faces against a small set of people you choose.
Scraped the whole of Zameen.pk for Karachi and predicts plot and flat prices. My first AI project.
Detects the emotion in a photo and adds matching tracks to Spotify, automatically.
Turns a script into a finished video with human speech and matching backgrounds, end to end, before generative video existed.
An HTML-to-JSX converter, a React Builder, an Atom Builder, and Python scripts that generate whole admin modules.
An auto-bidding bot for Freelancer.com, a LinkedIn jobs scraper, an email extractor and WeChat automation. If I do it twice, a bot does it the third time.
Cleaning, analysis and charts of the 12-13-2020.csv COVID-19 dataset. My first data project and my first repo on GitHub.
Messenger, Instagram, a Wattpad-style bookstore, a CS:GO skins marketplace and more, rebuilt in PHP just to see how they worked.
A console Ludo game on a 15×15 array, first semester of BSCS. Seven sleepless nights, zero idea what I was doing. It worked.