Apart from entrepreneurship, I have taught different people many things since 2005.
I used to teach Maths to class 12th students. Later I taught Marketing and Sales to MBA students, then I taught Digital Marketing to working professionals and also taught Entrepreneurship to new entrepreneurs. So you can say I have a flair for teaching.
While learning, my teaching instincts were also active and I realised that many AI concepts can be simplified in simple language.
It is an observation and many people (even coders) fear learning AI. They have a lot of misconceptions about AI
However, AI is just another technology and if we explain AI concepts in simple language, then we can definitely eliminate this fear of learning AI.
AI Concepts in Simple Language
With this thought process, I started a blog. My intention was to simplify AI concepts. I did write a few articles simplifying AI concepts and presenting them in a very simple language.
I also shared my learnings on LinkedIn and maybe seeing those posts, I got a call from a few engineering colleges (like JNU Delhi, NIT Delhi, University of Guangdong, CTAE – Udaipur) to take a guest lecture for their students on Large Language Models.
This was around August’2023 and during this period, I also got an opportunity to teach AI to students of Geetanjali Institute of Technical Studies, Udaipur.
As there was a semester break, the management wanted to train their Computer Science students on AI and we had a month’s time. They were also keen on covering Large language models as well starting from Machine Learning.
So, keeping these requirements in mind, I created a 20-day class schedule for them, starting from Basics of Machine Learning to Large Language Models.
This was the first time I taught my own designed course.
Genesis of AI course for IT professionals
Once that course at GITS was completed, SparkleHood (a community of the Top 2% professionals of in India) invited me to give a talk on Large Language Models.
It was surprising for me to see that even senior professionals and most of whom are non-technical are keen on learning LLMs. This was definitely the case sometime back.
After the presentation, when I interacted with a few folks, I realised their intention for learning was different and their questions were different from engineering students.
Basically, there is a new segment of learners out there that wants to learn AI and has different needs.
So, maybe the approach also needs to be different.
Problem – AI education for working professionals.
Once I realised that senior professionals are also looking to learn AI, I did a quick research and found a handful of courses for working professionals.
In many of the courses for working professionals, the curriculum is a copy-paste of the curriculum for students.
This is understandable because the majority of all existing courses were created keeping students and junior professionals in mind as they were the only learners before the ChatGPT era.
However, we all know that the needs of Senior professionals are different than the needs of Students.
Students need to go deep, they need to do more hands-on and master the technology while working professionals need to go wide.
They need to understand the possibilities of AI, how to build an AI team, the costs involved, real use cases of AI in businesses etc.
Well, coming back to the courses, these courses are offered by prestigious institutes like IITs, IIMs and top-notch universities in the US.
There are courses from reputed online ed-tech platforms as well.
The major problems with existing courses are
Spend so much time teaching Maths and Statistics
Do not cover the latest topics like Large Language Models, Generative AI, AI Agents, Multi-Modality, Federated Learning, Advanced Retrieval Techniques etc.
Long duration: 6 months to 1 year
Do not cover management topics like – AI/ LLM Security, How to build AI teams, Deployment, etc.
This led my entrepreneurial instincts to take over Teacher in me and I smelled a problem/ gap in the market.
Approach for building relevant AI courses for working IT professionals.
So, being an IT entrepreneur/ professional myself, I decided to solve this problem and started working on a dedicated course.
As I was working on the course, I realised that the objective of learning determines the topics.
Earlier, AI and ML were only about building models. The overall task is to gather data, do data engineering, build models, validate models, finalize them and deploy them.
But now with Transfer learning and open-source models readily available, not everyone has to build AI models. In fact the businesses/ organizations are either using proprietary models like OpenAI’s ChatGPT or open-sourced models.
Currently, there are 2 objectives to learn AI. If you want to become
1.AI Scientist/ Researcher –
The role of an AI scientist is to research and develop new AI models. The role requires a deep understanding of AI concepts.
You need to be very good in Maths, Stats and Python. Either you could do a PhD or take a comprehensive course available online
2.AI Product Manager or Senior AI Leader –
In this role, you do not have to build AI models but use AI existing models and deploy them for business use cases.
Therefore, there is no need to learn Maths and Stats. You also do not need to learn many Computer Vision and NLP concepts. Therefore, it considerably reduces the course curriculum and also the duration of the course.
Solution – Practical AI + Gen AI Course
Keeping learners’ 2nd objective of becoming AI Product Managers or Senior AI Leaders in mind, I started working on the first course
My thought process for teaching how to build models is to quickly show how to build and validate models through code rather than explaining the whole mathematical process of partial differential equations.
While other AI courses, talk about lots of Mathematical concepts like Probability, Partial Differentiation, Matrix Multiplication, algebra etc., I tried to give just a reference of equations of line and polynomial equations.
Once the learner knows about Pandas, I started with Supervised Learning and building models using the SciKit Learn library. The idea is to build 1 model end-to-end from scratch.
Most people built Kaggle’s Titanic model as their first model, which I also added to the course. Once the learner solves the Titanic competition, he/she gets an understanding of traditional ML models.
We get here in just 5 sessions, while other courses take 5-to 6 weeks and sometimes months.
After Supervising learning, it’s time to learn other ML techniques like UnSupervised Learning, Reinforcement learning, Semi-Supervised and Self-supervised learning.
With this, we are done with traditional ML and it’s time to move to the fancy stuff – Deep Learning
In Deep learning, we start with Neural Networks, then cover Computer Vision, and NLP and by the end of session 11 we are done with all the important concepts.
In terms of practical applications, we build Image classification applications, Text-to-voice, and Video generation tools.
We then start with Generative AI and Large language models and spend a considerable amount of time till Session 23.
Now comes the management part of AI where we learn about AI security, Regulations in AI, Hardware for AI, Deploying AI/LLM applications on the cloud, and real-life use cases of AI in business and then we add the course with the latest happenings in AI and LLM world
In this article, I tried to cover the discovery of the problem, my thought process and my approach to solving the problem.
In the next article, I will cover what exact sub-topics I included and more importantly what topics I did not include as the clarity comes not only from what to do but also from what not to do.
I actually removed more than 80% of the stuff which are not relevant for senior IT professionals.
I will cover everything in the next article.
In case you want to learn AI + Gen AI through instructor-led coaching, register for a free demo now.
Post Views:264
Share it with your senior IT friends and colleagues
