Adnaan Khan — I help AI leave the demo and get to work.
hi, I’mAdnaan.
I help AI leave the demo and get to work.Follow the rabbit hole
Right now
Forward deployed at KPMG AI LabOpenAI certified FDE · TorontoFirst, I watch
how you work.
Workshops, process maps and a lot of “wait, why do you do it like that?”
Then I build
the thing.
Prototypes in days, not quarters. Retrieval, orchestration and prompts wired into the actual workflow.
- Discover
- Prototype
- Evaluate
- Ship
- Adopt
Then I prove
it works.
If it isn’t measured, it isn’t ready.
- Accuracy
- Groundedness
- Hallucinations
- Retrieval quality
- Latency
- Cost
- Safety
- User feedback
Then I teach
everyone else.
On stages, in workshops and in my newsletter. Latest talk: “How to build with AI without creating slop.”
That’s the
whole job.
you reached the bottom of the rabbit hole.Read the long version Say hello
Forward deployed, for real.
I sit with the people who do the work, find the parts AI can genuinely help with, build it with them, and stay until it’s part of their week. Not a slide deck. Not a demo that dies in a folder.
- 01
Find the real problem
Most AI projects start as a vague “we should do something with AI.” I run the discovery workshops, map the workflow and turn it into scoped use cases with a value case behind them.
- 02
Prototype fast
A working prototype in front of real stakeholders beats a month of meetings. It’s the quickest way to find out what’s feasible and what people actually want.
- 03
Build it properly
Model selection, prompt design, retrieval, vector search and orchestration, set up so it holds up inside a real business process, with clear bars for reliability and performance.
- 04
Prove it works
Evals before opinions: accuracy, groundedness, hallucinations, retrieval quality, latency, cost, safety and what users say. That’s how a prototype earns the right to ship.
- 05
Make it stick
Enablement, change management, monitoring and feedback loops. I turn recurring tasks into reusable AI skills that non-technical teams can use, with a human checking the work.
- 06
Leave a playbook
Every engagement becomes an accelerator or a delivery playbook, so the next team starts halfway there. I also help leaders think through feasibility, governance, privacy and security.
What I actually wire up.
Every engagement is different, but the shape is usually this: AI plugged into the way a team already works, measured properly, with a person checking the output before it counts.
AI that works for people who don’t code.
I turn recurring tasks into skills with plain-language steps and a human checkpoint, so teams can use AI every day without an engineer in the room. Press run to see the shape of one.
An illustrative example, not a client project.
- Read this week’s meeting notes
- Draft a one-page update
- Flag risks and open decisions
- You review before anything is sent
KPMG AI Lab
Senior Consultant. Leading enterprise AI engagements from discovery to adoption, plus AI Lab initiatives beyond client work.
Oracle
DevOps and cloud engineering. Where I learned how software really runs in production, which is most of the job.
OpenAI Forward Deployed Engineer
The official stamp on the thing I already do every day.
I talk with my hands. A lot.
Talks, panels and hands-on workshops about building with AI without making slop. If your event needs someone who’s actually shipped this stuff, I’m happy to come by.
Invite me to speak
Today is not about becoming a “prompt wizard.” It is about becoming a better builder.
From my Toronto Tech Week talk
I write it down, too.
Got an AI idea that’s stuck in a deck?
Or a talk, a workshop, a role, or just a good question. I read everything.