Production AI & Systems Thinking
From claim-level provenance to AI operating systems — the architecture behind what I build.
The Halaqa is 1,400 Years Old. I Just Put One in Islam & AI
Nobody learned this deen alone — you sat in a circle. Islam & AI's new Halaqa Groups: private, invite-only study circles where every question and referenced answer posts to a shared thread.
Read →Six Weeks. One Guy. Six Things Live in the Market.
Six things went from dead to live in six weeks — ISNAD became a company, Islam & AI got paid tiers and mobile apps, and BunBites.pk got a real POS. One person, under $100/month.
Read →Nobody Knows Who Told Your AI That
Five agents touched the answer. If it's wrong, which one lied? Nobody knows — so ISNAD grades the transmission, not the truth, using a 1,200-year-old discipline.
Read →I Shipped 37 Commits in 20 Hours Without Opening a File
Zero-touch shipping is real — but the version of the story that stops at the commit count is a lie of omission. Three of the binary's claims were wrong, and the tests never looked.
Read →How a Solo Developer from Pakistan Landed on LangChain's Middleware Integrations List — Alongside OpenAI and Anthropic
A pull request, a 1,200-year-old methodology, and what the milestone actually means. ISNAD now sits in LangChain's middleware directory alongside OpenAI and Anthropic.
Read →I Built an Islamic AI That Refuses to Quote Scripture
The hadith scholars solved AI's hallucination problem in the ninth century. How Islam & AI is built to refuse to fabricate scripture — references instead of generated Arabic, no rulings, and disagreements shown rather than hidden.
Read →The AI Act Got Delayed. The Work Didn't
The high-risk deadlines slid to 2027 — Article 50 didn't, and the machine-readable marking deadline is fourteen weeks out. Why the standards gap is the real story, and how ISNAD answers the question that matters.
Read →I Graded 577,024 Hadith Chains to Find Out If My Library Was Lying
The benchmark I built to validate ISNAD — Cohen's κ 0.871 against scholarly consensus, negative controls, and the 1,200-year-old default that was right when my intuition was wrong.
Read →RAG Is Search. I Needed a Knowledge Base Instead
Why I stopped retrieving chunks at query time, and what I do instead — compile once at ingest, let contradictions get settled at write time, and make every page a file a human can open.
Read →I Have No Lab, No Press Office, and No Co-authors
The distribution system I built instead — a case study in getting independent research read, including the parts that didn't work.
Read →Five AI Agents Agreed and the Answer Was Still Wrong
Three chains, five transmitters, all agreeing. Except they all trace back to one NOAA report.
Read →Verify the Claim, Not Just the Agent: Inside ISNAD
The full engineering walkthrough — architecture, a worked example, the evaluation (including what failed), and why the hadith scholars were right about transmission by meaning.
Read →AI Has a Trust Problem. Muslim Scholars Solved It 1,200 Years Ago
How I adapted isnād–rijāl methodology from classical hadith science into a claim-level provenance framework — and why I'm building it in public.
Read →Can a 1,200-Year-Old Trust System Help Build More Reliable AI?
A classical idea from hadith scholarship inspired an open-source framework for tracing trust through multi-agent AI systems.
Read →Every Agency Founder Eventually Becomes the Operating System
I spent 90 minutes every Monday piecing together what was happening across 6 tools. The client I lost — the signals were all there.
Read →Most Engineers Are Preparing for the Wrong Interviews
At mid-to-senior levels, interviews stop being about knowledge and become about thinking — problem decomposition, trade-offs, handling uncertainty.
Read →The Gap Nobody Talks About in AI Careers
Most people aren't stuck because they don't know enough. They're stuck because they've never built anything that looks like real work.
Read →Visualizing Global User Distribution on Your React Website
Extract Google Analytics data and render real-time world maps with react-simple-maps — built for Islam & AI's 143k+ users.
Read →Career Leverage & Positioning
The mechanics of landing global roles — positioning, negotiation, and the numbers game.
The $200 Freelance Project That Changed My Career
A vehicle-detection Upwork project, a demo video posted on LinkedIn, 70,000+ views, six interviews at Motive, and a negotiated offer — from $200 to a career.
Read →How to Level Up as an Engineer: My Journey and Tips
From FAST-NUCES with a 2.69 GPA, to freelancing, corporate, startups, web3 — the full arc with lessons at every stage.
Read →How I Went From 95,000 PKR/Month to a $120,000 Remote Tech Job
The tech industry is not one market — it's a collection of completely different markets. Once you understand this, your entire career strategy changes.
Read →The Phase Where I Applied to 60 Jobs a Day
Not per week. Per day. When you're under financial pressure, you stop romanticising work. You just start doing what needs to be done.
Read →I Worked for Two Months Without Salary
What equity-only taught me about startups, cash flow, and survival. "You cannot pay rent with equity."
Read →The Job Offer I Rejected on the Day I Graduated
A retention clause that felt wrong, a decision that looked stupid, and a detour that shaped the entire career.
Read →Lessons from the Build
The stories behind the systems — freelancing, failure, cars, and the strange detours.
The AI Model Worked. The Startup Still Failed.
In 2021, I left a stable role to build a computer vision system for every mall in Saudi Arabia. The model worked. The system never reached deployment.
Read →Fiverr Banned My Freelance Career. Twice.
Years of work disappeared overnight — 350+ reviews, all gone. The lesson changed how I think about platforms forever.
Read →The First Car I Bought With Freelancing Money
Three weeks of $40–$200 coding projects, a 1980 Datsun 120Y, and my friend Umar at the Sunday Car Bazaar. Meet Paro.
Read →You Never Know Who Is Quietly Watching Your Journey
When I first started posting online, I believed nobody cared. The real impact is invisible — likes and comments miss it.
Read →MUST HAVE Portfolio Website for $3
Building a portfolio that costs less than a coffee — and works harder than most paid ones.
Read →