AI GOVERNANCE · ADVISORY · TRAINING

Your AI isn’t lying. Its map is wrong.

I find the small fix underneath your AI — the data it reads, the rules it follows, the guardrails around it — that puts your business back on the right road.

Video: a car's GPS says turn right, straight toward a lake. A sticky note on the windscreen corrects the route, and the car crosses a bridge and arrives in the city.

AI-generated video
WHERE AI GOES OFF THE ROAD · AI GOVERNANCE & DATA

Everyone is building AI. Few are checking where it’s headed.

CONFIDENT. AND WRONG.

The AI reads data nobody explained to it.

A column called cust_flg means something to your team and nothing to a model. I write the context back into the data — table and field descriptions built for AI — so answers come out right the first time.

Small fix: describe the data. Big result: answers you can trust.

GREAT DEMO. THEN SILENCE.

The pilot works. Production doesn’t.

Notebooks don’t have customers. I take AI from demo to dependable: clear ownership after launch, evaluation that catches drift (including LLM-as-judge), and the boring plumbing that keeps it running.

Small fix: measure it. Big result: it keeps working on Monday.

NOBODY WILL SIGN OFF.

Legal is nervous. Rightly.

Responsible AI isn’t a slide at the end. I build governance into the work — risk, bias, privacy and accountability checks — so the people who must approve it can say yes with a straight face. I work with US and Canadian AI governance frameworks, and teach how the US, the EU, Canada, India and China each regulate AI.

Small fix: govern from day one. Big result: it actually ships.

HOW I WORK · THE AI MAP CHECK

The map check.

one small tap, a whole new route ↓

  1. Read the map

    Before anyone builds anything, I look underneath: the data your AI reads, the decisions it supports, and the rules it has to follow.

  2. Make the small fix

    Usually it’s not a new model. It’s one missing description, one unowned process, one guardrail nobody added. I find the change with the biggest effect.

  3. Keep it on course

    Evaluation, monitoring and a team that knows how to drive it. I train your people so the fix outlasts my contract.

SERVICES · AI NAVIGATOR · AI MAP CHECK · AI TRAINING

Pick your route.

THE MAIN ROUTE

AI navigator

A few hours a month of senior advice. You drive; I read the map. Your team owns delivery, and I make sure the route avoids the lake.

  • Remote, a few hours a month
  • Second opinion on AI plans, vendors and designs
  • Governance and evaluation in from day one
Plan a route (opens Google booking page in a new tab)

START HERE

AI map check

A focused review of one AI system or plan. You get a plain-language report: what’s working, what will drive you into the lake, and the smallest fixes that matter most.

  • Data readiness and context for AI
  • Governance and risk review
  • Prioritized fix list
Book a map check (opens Google booking page in a new tab)

FOR TEAMS AND UNIVERSITIES

Training & workshops

I’ve taught AI at universities for about five years and trained thousands of educators before that. I teach the way I build: practical, honest about limits, and occasionally funny.

  • AI literacy for leaders
  • Prompting, agents and evaluation for builders
  • AI governance across the US, EU, Canada, India and China
Plan a session (opens Google booking page in a new tab)
SELECTED WORK · AI EVALUATION · AI-READY DATA

Small fixes. Big roads.

OPEN SOURCE

postline — LinkedIn, inside Claude

A free, MIT-licensed connector that lets Claude draft, schedule and publish LinkedIn posts — built on the official API only, and nothing publishes without a human saying yes.

View on GitHub →

EVALUATION

Catching confident wrong answers

An LLM-as-judge evaluation harness that scores AI answers before customers see them — so quality is measured, not hoped for.

Regulated-industry production work, described generally.

AI-READY DATA

Teaching AI to read the map

A hands-on training on writing table and column descriptions that AI can actually use for text-to-SQL — built with synthetic data, so any team can try it.

Repository coming soon.

WHAT PEOPLE SAY · COLLEAGUES AND STUDENTS

Don’t take my word for it.

Arooj is a standout data science professional with a blend of strong technical expertise and exceptional communication skills. She can transform complex concepts into clear explanations … [and] rapidly transitions from novice to trusted expert on new topics.

C. Victoria McKeownHer former senior manager · now Lead Data Scientist, AI and Innovation, McKesson Canada

The sessions were interactive and focused on real-world examples, making complex topics easy to understand and apply. Arooj always guided us on how to approach and solve practical problems.

Vikas DomakuntiPrincipal Software Engineer, Walmart Global Tech · former student, Post Graduate Program in AI/ML

Not only does she possess deep expertise and extensive experience in the field, but her natural ability to explain complex concepts in the simplest, most relatable way truly sets her apart.

Azmat MerchantProduct Owner, BI and Analytics Manager, Cencora · student, UT Austin AI/ML program

Her role went beyond that of a normal instructor; she became a mentor, bettering all of us in the process.

Justin ButlerInventory Analyst, Farm Boy · former student, Carleton University boot camp

All recommendations on LinkedIn →

WRITING · ARTICLES ON AI

from my notebook ↓

Analogies, honest results and the occasional existential crisis on a plate.

Read everything on LinkedIn →

ABOUT AROOJ QURESHI

Teacher first. Engineer always.

I spent ten years in education in Pakistan, leading training for more than 10,000 school leaders and teachers. Then I moved to Canada and into engineering — an M.Eng. at McMaster, a data science certificate from the University of Edinburgh — and have shipped production AI in telecom and energy ever since. Read the story: From Pakistan to Canada →

Today I lead data and AI engineering, teach AI at universities, and take a small number of advisory clients through AAQ Corporation. The thread through all of it: AI only matters if it works in the real business, for real people, the right way.

ask me about peanut butter and prompt engineering.

Arooj Qureshi, smiling, in a black-and-white portrait
  • BASED INHamilton, Ontario
  • EDUCATIONM.Eng., McMaster · MPhil, Leadership & Management
  • CERTIFIEDCPMAI · AIGP in progress
  • BUILT WITHRAG & GraphRAG, LangGraph, Vertex AI, agentic systems
FAQ · WORKING WITH ME

Questions, answered.

How do I get in touch?

Email me at hello@aroojqureshi.com, or book a free 30-minute call with any “Book a map check” button on this page. Tell me in a few lines where your AI is today and what worries you about it. I read and answer every message myself.

How do I book a meeting?

Click “Book a map check” and pick an open time on my booking page. The call is free, 30 minutes, on Google Meet, and the invite arrives by email. In that call you describe the road you’re on, and I tell you honestly whether I can help.

What does an AI navigator actually do?

I advise; your team drives. A few hours a month, I review your AI plans and designs before they ship, help you decide what not to build, check vendors’ claims, and make sure governance and evaluation are in from day one. Decisions and delivery stay with your people. You get senior judgment on call, without a new hire.

Do I need a navigator or a full-time hire?

If AI is one of several priorities and your team needs a second opinion more than a new boss, start with a navigator. It’s faster to begin and easier to adjust. When AI becomes core to the business, hire a full-time AI lead; I can help you write the brief.

What does an AI map check include?

I review one AI system or plan: the data it reads and whether that data is described well enough for AI to use, how its answers are checked, and the risks you would have to explain to legal or a regulator. You get a plain-language report and a ranked list of the smallest fixes that matter most.

Do you train employees in AI governance?

Yes. I run workshops for companies and associations on what AI governance means in day-to-day work: what to check before AI touches customers, money or personal data, who owns each decision, and how the US, the EU, Canada, India and China regulate AI. Each session is built around your tools and the decisions your people actually make. I deliver in person anywhere in Ontario, with travel included, and online for teams elsewhere on request.

Can you train forward deployed engineers?

Yes, and this is where governance matters most. Forward deployed engineers work right beside the customer and have to find the problem and ship a fix fast. Under that time pressure, checks get skipped. I train them to build governance in as they go: what data they may touch, how to test an answer before the customer sees it, and when to stop and escalate. Moving fast shouldn’t mean cutting corners.

Do you run AI workshops for schools?

Yes. Before engineering, I spent ten years in education and trained more than 10,000 school heads and teachers. I run practical AI workshops for independent schools and school boards: what AI can and can’t do for teachers and school leaders, and how to use it responsibly with student information.

Can you build AI courses for our continuing education program?

Yes. I design and teach AI courses for university continuing and extended education. I designed the introductory course of the University of Manitoba’s AI micro-certificate and taught it three years running, and I teach AI at the University of Toronto and Carleton. I can develop a new course or certificate around the gap in your program.

Do you offer personal AI classes?

Yes. For individuals, I teach personalized AI courses through GetInTech, my own course platform.

What is AI governance, and when do I need it?

AI governance is the set of rules, checks and owners that keep AI legal, fair and accountable. You need it as soon as AI touches customers, money or personal data. I work with US and Canadian frameworks and teach how the US, the EU, Canada, India and China each regulate AI.

Where do you work?

I’m based in Hamilton, Ontario, and work remotely with clients across Canada and the United States, with on-site visits when a project needs them.

Is your AI heading for a lake? Let’s check the map.

A free 30-minute call. You describe the road you’re on, I tell you honestly whether I can help.

How I handle what you send me: privacy notice.