Once upon a time, a book on human anatomy mesmerized a little 3rd-grade girl on how a human brain works. For a young girl, though it was difficult to understand the real functionality of the brain, her mind was sharp enough to appreciate how the complex web of millions of biological neurons works together to produce human actions. Years later during her engineering days, she stumbled upon artificial neural networks of machine learning and it rekindled her childhood curiosity with the human brain. She not only fall in love with neural networks but also dwelled herself into the world of machine learning. Meet Priyanka Kasture, that little girl who now owns the largest machine learning community on Instagram.
Priyanka is an ML influencer and the founder of Instagram page Machine Learning India (MLI) which is now very close to cross 200,000 followers. Even her Telegram channel and LinkedIn page have 30,000 plus followers each. It is not the sheer size of the followers which makes it so impressive but it is how Priyanka shares ML knowledge to the community by sharing the best possible content is what makes it stand out.
You would not find any memes or copied content on the MLI page, every post and story is created and designed by Priyanka. What makes her content so successful is that they are in accordance with what her audience demands. She has also organized several physical workshops on machine learning in multiple cities in India. MLI also collaborates with other AI startups and personalities to give the curated ML content to its community.
The LinkedIn profile of Priyanka mentions she is a growth hacker and a little more research will reveal that she has entrepreneurial streak as well with a couple of small ventures during college days. So now it makes sense that how an ML geek like her could create a community brand out of an Instagram page.
Interview
We got an opportunity to catch up with Priyanka to understand her journey with machine learning and MLI. We spoke about the current landscape of artificial intelligence, future AI trends, and how beginners should approach machine learning. Below are the excerpts of our engaging interview.
MLK – Priyanka thank you for giving us this opportunity to have a conversation with you. So the first thing we would like to know what brought you to do the field of machine learning?
Priyanka – Yes, so in 3rd class, I came across a book that discussed the brain and I was amazed at how the brain has so many different parts and it processes all sorts of information. So this drove me to find more about the field, but as I was growing up we got into 10th,11th, 12th then JEE preparation and stuff like that. So I had to leave all my hobbies and interest aside to prepare for these exams.
But when I came to my engineering I was experimenting a lot in my 1st year and I came across an amazing video on YouTube which mentioned AI and as well as Neural Networks, and it took me back to my childhood and again triggered my interest with the brain and neural networks and how they function. That is how it all started I would say, I started exploring the field, took up the neural network course by Prof. S Sengupta from IIT Kharagpur under NPTEL on YouTube. Of course, it took me a long time to complete this course as I was still naive. After this, I realized that I should have finished machine learning first as that is considered a step prior to study neural networks (deep learning). Then I started taking up courses on machine learning, Python, and did plenty of projects.
MLK – What were the main challenges you faced in your initial days of learning ML?
Priyanka – Here if you see, I made plenty of mistakes as a beginner, for example, I started with the neural network before the machine learning. I did not study Python before I started with the neural network or machine learning. So I felt the lack or absence of a guide and mentor throughout my journey. Apart from this, a couple of other challenges were that back in 2015 we did not have a lot of mature platforms (for machine learning), Tensorflow used to break down on windows, there was no proper documentation, we did not have good answers on Stack Overflow as things were very very new back then and so learning was very difficult. Moreover, (since the curriculum was very different), I had to self learn a lot of things without any help from college through online courses.
MLK – Could you let us know what were the books and courses you studied so that it can also help our readers?
Priyanka – The course on Coursera by Andrew NG is amazing but it is Matlab right, so for Python, you can take up Python Masterclass on Udemy. For the practical aspect of machine learning, you can take Machine Learning A-Z by Superdatascience team. All these courses were available back in 2015-2016 when I took them, but now we have better courses available. So beginners can experiment with a couple of courses and then settle on one particular course, there is definitely not one best course out there. For books, I will recommend Elements of Statistical Learning by Trevor Hastie from Springer publication, it is a great book for beginners.
MLK – Coming to your Instagram page, Machine Learning India (MLI), what was the intention behind making this page, and how had the journey been like till now?
Priyanka – As I told you, that I felt the lack of guide and mentor in my early days of machine learning and I did not want other beginners to go through the same struggle and this is why I created the Instagram page Machine Learning India. When I started with this page, obviously it was a bit difficult for me as I did not know about how social media content and designing works but people started loving my posts.
Machine Learning, Data Science, AI was a niche back then as there were not many Instagram pages that were posting quality information. So it worked in my favor as I was one of the few pages that was giving out quality information with amazing design. It took me 10 months a follower base of 10K and after one year it was near about 30K and now it has been two years and now we have 200K plus people on multiple platforms. We are now adding 25-30K people per month. So I hope it will 300K-400K people by the end of this year looking at our content.
MLK – How did you make sure MLI stands out from other ML pages on Instagram?
Priyanka – Here what happens is that unless you are original you are not going to grow, you will always remain the shadow of the person you are copying. This is what creators need to understand that people will not like to follow 10 pages that post the same content. Instead, they will follow only 2-3 pages that post good and original content. So I will say that good, original, and crisp content with an amazing design that generates a good amount of engagement and bond with the audience – was the recipe to grow our Instagram community.
MLK – When did you realize that MLI is moving towards something big?
Priyanka – So that was way back when the page got 5000 followers because I never had the idea to make the community out of it. Only after it reached 5000-6000 followers I started calling it a community.
MLK – There are too much hype and misinformation that is floating around with respect to AI, what is your opinion on that?
Priyanka – Whether AI is going to take over or take your jobs or stuff like these, see media sensationalizes AI really which creates panic among people in our community and beyond. They say AI will take your job but the fact is AI is struggling and not maturing that well. The complex deep learning systems that we have today in the biggest organizations, all of these are simple pattern recognition systems, they are just black boxes and explainability is a big question here. AI is still struggling in reasoning and this technology will take more time to mature.
But having said this, those laborious jobs that do not require creativity can easily be automated. In fact, a lot of parts of data science and machine learning pipeline is itself getting automated. In years to come, we will have automated machine learning and our job will only be to monitor them. But again if we emphasize on problem-solving, reasoning, and logical thinking that cannot be automated, only humans can do that.
MLK – We had at least two AI winters in past, do you think right now this is an AI bubble that will burst soon or do you think this AI boom is here to stay?
Priyanka – There may not be an exponential growth in the future, but AI will going to stay. This is for a very simple reason that data is getting generated at a huge volume and with time we are going to produce more and more data. So we really need systems that can derive insights from this data, insights that cannot be easily seen by humans and we have AI systems that can do that for us. These insights are very important as they help in marketing, finance, they help us make quality decisions for businesses.
MLK – In the next 5 years, what trends do you see in terms of the AI job market. Do you think it will reach a saturation point?
Priyanka – Machine Learning can become saturated for a very simple reason that many people are taking up courses and there will be a time where you don’t need more machine learning engineers. But if you take a step ahead from simple predictive modeling and try to specialize in NLP, Computer Vision, Robotics, reasoning systems, or deep learning or whatever it is, if you try to specialize in these then there is definitely a lot of scope in the future.
MLK – Do you think Covid-19 has put the AI job market at risk because companies are refraining to invest in business transformation currently and what should people do at this time?
Priyanka – Well, I will not say that only AI job market as such, but if you take any technology sector there is going to be a hiring freeze for the next 3-4 months. So what is important for people to understand here, especially students who are looking for jobs, that this is a good time for them to build their portfolio. Mid and large companies are not hiring but some startups are hiring interns and you can try for those else you can make your own project portfolios, contribute to open source projects, participate in online hackathons. So once lockdown ends and companies start hiring again you should be ready for the job market.
MLK – We would like to know, whom do you look up to as an inspiration in our AI community?
Priyanka – Well in the AI community there are few people, one is Professor Andrew Ng himself who has done amazing work. He started as the professor in Standford, then he was working with Google and now he is with Baidu. Apart from him, Abhishek Thakur, he is 4x Kaggle Grandmaster, who could be a better inspiration right. Then Francois Chollet who is the creator of Keras has also done very inspiring work.
MLK – We know many might have asked you this question but it will be great if you can let our readers know, how to start the journey of ML and data science?
Priyanka – First of all, this will really occupy a good amount of time from your life so don’t pursue machine learning only because it is hype, everyone is doing it and you think it is cool. Not everyone has to become a data scientist or machine learning engineer. You could be an amazing web developer, web designer, you could be in cybersecurity.
Secondly, there is no perfect time to start, there are many who come and ask me that I don’t think I am good at Python or maths or problem-solving so should I get started. So I want to tell here is that situation is never going to be right, you just need to hit the ground running and get started. And if you are really passionate about machine learning you will figure out the way, just like I did when I learned deep learning in 1st year in spite of not knowing coding or calculus.
So as a beginner place more emphasis on problem-solving, critical thinking, business domain understanding. We see so many people attending machine learning conferences but how many of us attend conferences of marketing, finance, or banking or any other industry. Machine learning is just a tool that you apply to different domains so beginners should have some basic knowledge of that also.
And I have already mentioned some courses and books earlier for beginners but there is nothing like the perfect course, experiment with few and eliminate the ones which you don’t like.
MLK – 5 Mistakes which you see beginners are making when pursuing ML?
Priyanka – First of all, nowadays youngsters are so aware of machine learning that they start machine learning just like that from 1st year. But what is important for these guys to understand is that there are other core concepts like data structure, database, that you should also focus on initially. But people are jumping to machine learning without basic computer science concepts.
The second mistake is that you should also be aware of programming languages like Python before ML. People do not do that and now what is happening is that we are producing a very poor quality of data scientists who cannot program and only know how to import libraries that is at a very high level. If you go back 10 years from now, graduate students used to write neural networks in C, how many students can do that today.
The third major mistake that people do is that they jump into ML without knowing maths. Unless your basic understanding of linear algebra, probability, statistics, calculus it is not recommended to get into ML.
The fourth mistake which beginners do is that they do not build their project portfolio, they only take online courses and start expecting companies to give them jobs or internships. But unless if you cannot prove to them that you can build an end to end machine learning application they are not going to hire you.
The last mistake is not to build a presence on online platforms like LinkedIn or Github. Being good at machine learning or data science is one thing but if you cannot market your skills it is of no use. People around you and recruiters should know what you are doing in machine learning, personal branding is really important over here.
MLK – So what are your future plans with the Machine Learning India community?
Priyanka – For Machine Learning India, I will see it as a global community for machine learning and data science. In fact, very soon we are going to shift our focus from machine learning to deep tech. So we are going to cover domains like blockchain, quantum computing, and build a community around deep tech.
MLK – So what is your future plan over the next 5-10 years is it going to be around ML/AI or you have something else in your mind?
Priyanka – I plan to make my career in marketing so that is where you can see me. It will not be digital marketing as such, but you can say community management, online branding, content marketing, that is where I see myself 10 years down the line.
MLK – Thank you very much Priyanka for giving us the opportunity for this insightful conversation around the topic of machine learning.
Priyanka – Thank you.
