Pipelines, AI products, and the conversations that get them adopted. I've never had to say "let me loop in an engineer."
At a bank that meant the pipelines, the quality framework, the dashboards, and the cost strategy. On my own time, it's a voice AI toy phone.
I build the checks that catch bad data before a human has to. At a bank, that was 144 automatic checks watching 36 financial tables, 60% fewer data fires, and an audit trail that wrote itself.
I also built a tool that turns a plain-English question into a database query, so an answer doesn't have to wait for an engineer.
ARIA is my sales research robot. Give it a list of companies and it reads up on each one, spots who's about to buy, finds the right person to talk to, and drafts the first email, on a schedule, while keeping an eye on competitors. Go-to-market work, automated, so a sales team spends its time on the conversation instead of the research.
Present the findings, make the case, walk a room through the product until they see what I see. It starts with listening, because you sell when people feel understood.
Not a side skill. Half the job, and the half I enjoy most.
Most people pick a lane. I keep ending up in both. I build the pipeline, then I'm the one on the call explaining why it matters, taking the hard question, getting to yes.
Data engineer, solutions engineer, sales engineer: the industry has a name for every half of me. I just call it the job.
Pipelines, quality checks, backends, automations. The plumbing. I like boring reliability, because boring is what keeps the demo smooth and the pager quiet.
At Capitol Federal I was in front of marketing, digital banking, and small-business teams almost every day. One good chart and a plain sentence beat twenty slides. People decide faster when they understand.
Find out what someone needs, show them the product doing exactly that, and stay until it's part of their week. Not a salesman. A builder who likes the customer.
Home. Grew up here, started playing soccer at eight, and grew up bilingual. Urdu first, English second.
I moved to the US at twenty for college. Computer science at the University of Kansas: systems, algorithms, databases, and the discipline to build things that hold up. Everything I ship now started here.
The city I always wanted, and the one I'm building in now: data systems, AI products, and the conversations that get them adopted.
Clarify what's actually being asked. Look at the system before guessing. Say the trade-offs out loud, then commit. Thomas Sowell said it best: "There are no solutions. There are only trade-offs."
I'd rather ship something dependable and explain it well than ship something clever nobody trusts. I'm competitive, too. I just aim it at the problem, not the people.
Two jobs and one thing I'm building on my own. One taught me how a bank moves data. One taught me to fix things while the person watches. The third is teaching me everything else.
I sat between the engineers who moved the data and the teams who ran on it. Some weeks that meant pipelines, quality checks, and query tuning. Other weeks it meant turning what the data showed into strategy for marketing and digital banking: which customers to reach, in what order, and what to say to them.
I was asked to build cost monitoring. I expanded it into a full platform: six months of query history, warehouse cycle patterns, contract burn forecasts, and a scoring model for which warehouses to merge. It showed only 21 cents of every dollar doing real work, one always-on warehouse costing $5,918 a month, and 78% of spend sitting idle. The data architect cited it in the formal strategy: 21 warehouses down to 7, $74k to $96k a year.
A hundred-plus issues a day, most of them in person. Someone walks up with a deadline and a broken machine. You diagnose it fast, fix it, and explain it in a way that lands. I wrote the troubleshooting playbook that made the team about 20% faster.
This is where I learned I'm good in front of people. Also where I learned "have you tried restarting it" has a better hit rate than most models.
My niece kept asking a Peppa Pig toy phone real questions. It had nothing to say. Aleya is the answer: a warm, safe voice friend for ages 2 to 6. Hold the button, talk, she talks back. Pretend play, kid-sized answers, comfort, a second language when the kid switches, and a gentle "let's ask Mama" on anything she shouldn't touch.
Honest status: not a funded company, not a pitch deck. Something I'm building because it should exist, and the software works today. I own everything from how she thinks to what she's not allowed to say. The actual toy comes next.
Work and side projects, kept apart on purpose. Every one of these has come up in a conversation as "wait, how did you do that?"
Four dashboards the bank still opens every morning. 18,597 accounts studied, 1,452 small businesses found that nobody had ever called. Best finding: a new customer either sets up online banking in the first six days or never does. That changed the campaign.
144 automatic checks watching 36 financial tables, run from one rules table instead of a pile of scripts. Is the data fresh, complete, adding up to the cent, matching between layers? If not, the right person gets an email and the compliance team sees it on a dashboard. 60% fewer data fires.
Dug through 12.7 million queries and six months of bills. Found that only 21 cents of every dollar was doing real work, one machine left running around the clock was burning $5,918 a month, and 78% of the spend was idle. Wrote the case for going from 21 warehouses to 7: $74k to $96k a year. Nobody has missed the other 14.
A toy phone for kids that actually talks back. Listens, thinks, checks itself, answers in a warm voice, remembers names and favorites, and redirects the grown-up questions to parents. The plan is to bake her personality into the model itself.
A research robot for sales teams. Give it a list of companies and it reads up on each one, spots who's likely to buy, figures out who to talk to, and drafts the first email, on a schedule, while also keeping an eye on competitors. I built the machine a sales team wishes it had.
A city traffic simulation where the cars and the traffic lights teach themselves, by trial and error, how to keep things moving. Runs as a set of small services in the cloud, redeployed automatically every time the code changes. The cars learned to stop hitting each other. Eventually.
Who wins the Premier League this season? A database of every match, a pipeline that pulls fresh stats automatically, a forecasting model, and a dashboard where you can play what-if. Built by a fan, checked by numbers, occasionally wrong in ways the actual table would agree with.
Type a question in plain English, get the answer straight from the database, no SQL required. It got the query right 85% of the time across 50-plus test questions, and it keeps a log of everything it ran. One file for the prompt, one for the model, one for the database, none of them pretending to be the other.
A model that predicts which customers are about to leave. The interesting decision wasn't the model, it was the dial: I tuned it to catch 62% of leavers at the cost of some false alarms, then matched each risk level to a retention move that costs the right amount. Depends who's paying for the phone calls.
The depth is the point. On a call, I'm not the one who says "let me loop in an engineer." I am the engineer. I also speak human.
Moving data from where it lands to where it's useful, checking it on the way, and keeping it fast. I've been the on-call person for a bank's pipelines. It builds character.
Clean, modular code and a lot of debugging. Comfortable in SQL all day, R when it's the right tool, Excel when someone senior insists.
The services behind the screen: take a request, check it, do the work, store the result, and fail loudly instead of quietly lying.
Apps built on language models, speech in and out, models that forecast and classify, agents that learn by trial and error. I know when the simple model beats the fancy one, and I'll say so.
Version control, containers, automatic testing and deployment, scheduled jobs. The unglamorous part that makes demos boring in the best way.
Picking the right metric, studying how customers actually behave, testing ideas properly, and turning all of it into a chart and a sentence an executive can act on. Google Data Analytics certified, for what it's worth.
The whole picture in one place: where I'm from, school, work, the thing I'm building, and what I do when nobody's asking for a dashboard.
I build the pipelines and automations, then I'm the one who demos them, takes the technical objections, and makes sure the customer gets real value after the signature. The industry has several names for that person.
I've played soccer since I was eight. It's where my head goes quiet. Rugby came later and it's the loud side, the part that needs contact and chaos. Both, because I'm competitive and I like having somewhere to put it.
Arsenal is my club, which is how a fan's argument turned into a full Premier League prediction model.
When I'm not on a pitch, I make music. Beats and lyrics. It's how I journal.
Coffee in New York, a pipeline problem, an argument about Arsenal, or a question about the talking toy phone. All welcome. I reply fast.