Tanmay Bakshi

How often do you get to be on the cover of a magazine, and get a letter from the Prime Minister of Canada and from the Lieutenant Governor of Manitoba? Today, I got all 3 in one 🎉! So honoured to have been a recipient of the Life Mentor Award from the @CreativeFdnWPG!

Learning to build AI and learning to use it are different things. Building it needs technical foundations. Using it should get easier as the tools improve. If every new iPhone needed a course, we'd stop buying them. From my student Q&A at @itmskillsu.

Are we building better products, or better pitches for funding? When “AI-powered” gets approval, teams optimize for that story. We need to reward customer value and trust engineers to choose what actually helps. From my ABP INFOCOM Hyderabad chat with Soumyajit Dey.

Sometimes the hardest part of building with robots is figuring out what to do with them that someone will actually pay for. Plenty of things I’d love to make this one do. Getting it to earn its keep is another question :)

I love giving back to the community through conversations like these, exploring what AI means for the future with students from DBM India’s Udaan programme and ITM Skills University :) Loved my time in Bombay. Thank you for the warm hospitality!

Before treating low AI adoption as resistance to change, I'd ask: how much work are we expecting people to do just to make our tools useful? A tool has to save more effort than its imperfections create. From my ABP INFOCOM keynote in Hyderabad.

An AI agent asking for permission sounds reassuring... until another agent with no authority gives it. That happened in the OpenAI/Hugging Face incident. "Locally sensible", globally undesirable! Had a lot of fun discussing this and more with @YellowRobotXYZ on @ai_ranch.

For the longest time, I didn't find it valuable to add AI to my daily development workflow. That changed this year: it became a genuinely productive collaborator. Tomorrow at @ABPNews INFOCOM Hyderabad, I'll explain what shifted, what's hype & how leaders can tell the difference.

Excited to be Live in India this July — keynotes and conversations, from students to industry leaders. Different rooms, different AI questions. Underneath, one answer — more hopeful than hype or fear: it raises the ceiling on what we can do. More work worth doing, not less.

"Who's accountable when an AI agent makes a mistake?" Nine years after sharing on the @nasscom stage why AI augments people rather than replacing them, I'm back — keynoting Governing the Second Workforce at People Summit 2026, Bengaluru, July 14–15. That answer? Still stands :)

My pa, my hero — in two months, he went from healthy, to choosing MAID (Medical Assistance In Dying) after a brutal fight with stage 4 gastric cancer. It began December 24 with a worrying ultrasound; he couldn’t eat, lost weight rapidly, and spent weeks seeking tests and care from his family doctor who was away on vacation — waiting rooms, trying to convince healthcare professionals that he was suffering, unable to get timely medical attention, and being told he wasn’t “sick enough.” We then resorted to private IVs while we saw him suffering at home. After finally being admitted, he endured chemotherapy. On February 18, 2026, exhausted and in pain, knowing that the treatment isn’t going to make him better, he chose to say goodbye to the world. He is my champion — he nurtured my passion for technology, and quit his full time job so that he could homeschool me, touch the lives of hundreds of students as a self-employed tutor, and travel with me to every keynote. His last words were to keep going and finish what he couldn’t. He wanted others to have a chance at a cure so he decided to donate his body for research and to donate his organs to help other Canadians. No family should have to live through this horrific, heartbreaking experience; a man quietly suffering while being denied the care he needed during his countless visits to the ER.

I love ML & coding, but hate that a lot of ML code is repetitive, brittle, and hard to scale. JAX is the next-gen framework that makes ML coding fun again! Join me on my journey of learning JAX & Equinox by building a Handwriting Synthesis Transformer 👉 youtu.be/yF7sFG9VmVk

We taught Pepper the Robot to chat & extract answers from docs using watsonx.ai—all in just a few hours! At @BMO TechConnect 2024, I picked up where I left off in 2023, and we explored how Gen AI capabilities are rapidly advancing & becoming accessible to everyone.

Back in 2017, I flew a drone on-stage by talking to it using AI. In 2024, OpenAI used LLMs on-stage to write an app that controls a drone. AI progress sure is taking off! 🛫

The most successful companies in the future will be the ones that (finally) realize through this benchmark that their interviews aren’t effective at measuring the performance of their engineers, not the ones that use it as a reason to cut costs on people & blindly implement “AI”.

GDG Mississauga's "Build with #AI" event was a group of passionate people discussing interesting concepts like gradient descent, AI & #ML. Such events help kickstart students' & professionals' journeys into building real-world AI solutions—thanks for inviting me!