Where I come from
I have managed national-scale programmes, live product portfolios and cross-functional teams. Most of that work was operations in the unglamorous sense: building the dashboards, the standard procedures and the reporting that kept large moving things honest, and being the person accountable when they slipped.
At AGON Esports I run operations for CEGA Pakistan, a federally funded programme that had to go from nothing to live delivery across two cities. That meant intake funnels that processed thousands of applications, governance that had to survive an audit, and partners who all needed the same answer at the same time. At Glitch I manage mobile game delivery, and before that I spent years in game design tracking retention, monetisation and user-acquisition numbers daily across a portfolio of live titles.
None of that is automation work. All of it is the reason I am any good at automation work.
Why that order matters
Knowing which tool to use is easy, and getting easier every month. The hard parts are different questions entirely:
- Which fifteen minutes of someone's day is actually worth removing. The loudest complaint is rarely the biggest cost. The real drain is usually something nobody mentions because they have stopped noticing it.
- What breaks the week after launch. Automation fails on the edge cases — the duplicate order, the customer who replies three days later, the month with five Mondays. You only anticipate those if you have lived through them.
- What should stay human. A machine handling a complaint badly costs more than the salary it saved. Knowing where that line sits is judgement, not engineering.
The shortest version: I am not interested in showing you something clever. I am interested in the difference between what your team costs this month and what it costs next month, without anything getting worse.
How I work
01 / AUDIT
We find the expensive habit
A week of watching how work really moves through your business, not how the org chart says it does. You get a ranked list and a fixed quote — and if the answer is that nothing here is worth automating, the report says that instead.
02 / BUILD
One workflow, working properly
I build the highest-value item first and get it live. Narrow and finished beats broad and half-done, and you see the saving land before deciding to spend more.
03 / HAND OVER
Your team runs it without me
Written documentation, a walkthrough with whoever owns it, and 30 days of tuning. Accounts in your name. You are not locked into paying me forever to keep the lights on.
How I use AI, and where I do not
Language models are very good at a few specific things: reading a messy invoice, understanding an order written in Roman Urdu, drafting a reply that sounds like a person. They are unreliable at anything that has to be exactly right every single time.
So the split is deliberate. Judgement and language go to a model, with checks around it. Money, dates, stock counts and confirmations run on rules, because a model that is right 97% of the time is wrong once a month in a way that costs you a customer. When I build something, I will tell you which half is which.
The same applies to how I build. I direct AI coding tools heavily and I check everything they produce — the Miqyas case study lists the specific bugs that caught, including a data error found only by recomputing the totals by hand.
Working with me
- Fixed scope and fixed price, agreed before anything starts. No hourly billing.
- Everything in your name — the Meta account, the hosting, the documentation.
- A straight answer on the first call, including when the answer is that you do not need me.
I am based in Islamabad and work with businesses here and across the Gulf.