FRACTIONAL AI LEAD · AI GOVERNANCE · TRAINING

from my notebook ↓

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.

WHERE AI GOES OFF THE ROAD

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.

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

HOW I WORK

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.

WAYS TO WORK WITH ME

Pick your route.

THE MAIN ROUTE

Fractional AI lead

Senior AI leadership, part time. I set direction, unblock your team, choose what not to build, and own the path from idea to production — with governance in from the start.

  • Remote, with on-site visits when needed
  • AI strategy, roadmap and vendor choices
  • Delivery oversight and responsible-AI guardrails
Talk about a fractional role

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

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
  • Responsible AI in practice
Plan a session
SELECTED WORK

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.

WRITING

From my notebook.

Analogies, honest results and the occasional existential crisis on a plate. New posts land here when they go up on LinkedIn.

Read everything on LinkedIn →

ABOUT

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.

Today I lead data and AI engineering, teach AI at universities, and take a small number of fractional and 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.

  • BASED INHamilton, Ontario
  • EDUCATIONM.Eng., McMaster · MPhil, Leadership & Management
  • CERTIFIEDCPMAI · AIGP in progress
  • BUILT WITHRAG & GraphRAG, LangGraph, Vertex AI, agentic systems

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

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