Lesson series

Problem Definition in AI

A dream of Deep Learning is to solve complex multi-stage problems with a single architecture. While possible, it may often not make sense to do so. We show why & present ways of slicing & dicing a problem into smaller ones for solving and combining.

The Hero Methods Course (Free)



Starts at

Your convenience



Content Duration

90 minutes



Problem Definition is your weapon.

Better understand your domain, relate to your audience and lead to consensus.
Ignore it at your peril. We show you how to do it right!

#1 Hero vs. Non Hero

Unlike classic software builds, for problems in  AI, the solution is not always known. Much experimentation is needed to surface the answer.

#2 Common Ground

Regardless of which, you are never going to solve it right, if you don't know what you're solving. Problem Definition is the first thing you'll need to do.

#3 Maven of Modelling

Problem, not solution. You'll learn about it from stakeholders. You then decompose it into sub-problems - in different ways. The end goal - Modelling.

#4 Shepherd Stakeholders

Then, teach stakeholders the many dimensions of an AI problem. Show them how and why you decomposed yours, and get their feedback. Earn buy-in & consensus.
Best for Learning

Not the usual Machine Learning course!

Write your awesome label here.

Design Thinking

Ponder different ways in which a top-level problem can be structured, and solutions  woven together in an architecture
Write your awesome label here.


No hands-on exercises, Jupyter Notebooks or datasets here
Write your awesome label here.


Open-ended questions, discussions and deep thinking are best for learning, and are our primary method
Write your awesome label here.

Peer Learning

Interacting with peers and learning from their experience is better than exams, quizzes or certificates. 

Jagannath Rajagopal

I teach Design Thinking & Architecture for a whole universe of methods of which Machine Learning is but a part. Between high level use case content and hands-on coding, there exists a knowledge gap today. At Kado, my team of overachievers, and I have melted our brains and put down everything we know about Hero Methods towards a knowledge offering that is different from the usual run-of-the-mill Machine Learning course. Our goal is to bridge this gap.

I've wrestled Tensorflow to the ground, learned Python from scratch in the process, crashed and burned in 2 start-ups, launched a successful YouTube channel, failed at selling Scuba gear, taught R to many, and succeeded several times in implementing large scale Supply Chain Planning systems for big companies.
Write your awesome label here.
Created with