About GK
I'm GK. I help product teams raise their bar with AI.
I run three things at once: coaching product teams, AI consulting, and building an e-commerce brand. The UK recognised that range with a Global Talent Visa for Exceptional Talent in Technology. It is also why I teach AI from the product, marketing, and operating sides at once, not from theory.
What I do
I started teaching AI with Vital AI Academy in 2025, became a Maven Top Instructor soon after, and founded an e-commerce company in the UK along the way.
15,000+ PMs signed up to watch my free lesson featured by Lenny Rachitsky on Maven. Jason P. Yoong from Level Up and I co-teach live certification workshops on Maven.
I ran a workshop for Canva's product teams, training 100+ PMs to build with AI on first principles and to make AI and humans actually work together.
The reason I do this: I watched too many senior PMs, at Amazon and after, panic about falling behind their own team. They were asking the right question about AI, on teams that had no shared way to answer it yet. The gap between a good question and a team that can act on it is what I build the answer for.
How I got here
It comes from a decade that crossed Amazon product and operations, B2B SaaS marketing, and an MBA, before AI pulled them together.
Amazon
ESADE
Typeform
Chargebee
Zoho
+ more
Career highlights
- Senior Product Manager in a $7.2B business unit at Amazon, leading ML and econometric models that actually moved the needle on delivery experience
- Ops Management at an Amazon robotics site, leading a team of 300+, running 200K units a shift at 98.7% delivery accuracy
- MBA from ESADE (ranked #7 by FT): Social Impact Scholarship, VP of Tech & Media club
- Led marketing at B2B SaaS unicorns (Typeform, Chargebee, Zoho): inbound, virality, and demand generation
- Founded a non-profit in rural India at 19, for 5+ years, recognised with state-level awards
- Awarded the UK Global Talent Visa for Exceptional Talent in Technology
What I teach
I teach AI product work from first principles, the way I built it at Amazon and now teach it on Maven. Specifically:
- How to write a spec a model can build from, without a human translator.
- How to run an independent review calibrated to your taste, not the model's defaults.
- How to set a shared definition of "good," so it means the same thing to everyone on your team.
The myth I spend the most time on: becoming technical does not mean learning to code. It means writing logic that an engineer, or a model, can test against. PMs, engineers, and designers all learn it in the same workshop.
None of this depends on the model. When the next model drops, you do not start over: the spec still has to be clear, the review still has to hold, and your team still needs one definition of "good." The judgment behind those calls is the part worth building, and it carries to whatever ships next.
The team and what's coming
Motivated Code is a lean two-person team by design: I own the spec, Ricky owns the build. We work inside a client's actual workflow until it runs without us, then hand it over with the client in control. Where coaching teaches your team to build, this is where we build it for you.
What's coming: more Live Lessons, more Substack essays, more team workshops.
Three ways to work with me:
- Coaching, for teams and individual PMs.
- Motivated Code, where we build the workflow for you.
- The free resources, to start today.
Where it all began
At 19, I founded a child-safety non-profit in rural India and ran it for five years: awareness programs for 22,000+ children, 750+ abuse cases pursued with the district administration. Work like that runs entirely on trust, and you learn fast that it has to be built in from the start, not patched in after something has already gone wrong.
I bring that same instinct to AI. The executive assistant I built on Claude Code takes instructions only from the live session, stays sandboxed, and has a prompt-injection guardrail, so it cannot be talked into acting on something it reads. When I teach spec engineering, defining what a wrong answer looks like comes before the build, not after. It is a method I pressure-tested against Anthropic's published approach to building with Claude, and have taught ever since.
Whether I am training a team or building a workflow, the people on the other end, your team and their customers, have to be able to trust what ships. It was never optional for me.
Start with the free stuff.
Start with the free 30-minute lesson on Maven, or book a discovery call if you'd rather talk.