A passion for machine learning leads to Ideasinc success
By Sarvasv Kulpati, High School student, Dover Campus
In Grade 10, I decided to share my love of programming and creating with code with other students, and started an activity for coding at the IDEAS Hub. I was lucky to find that there were many resources available to help teach students programming, like drag and drop tools and simple explanations. As the year progressed, I decided to try and teach Machine Learning (ML) to the students, but encountered a problem. ML not only requires coding knowledge, it also requires an understanding of complicated math. I wasn't able to find a way to teach students how ML works without also having to go through advanced concepts that they might not be able to grasp.
Because of this, in February 2018, I started work on a project I now call MLBlocks. ML models are made of multiple layers of ‘artificial neurons’; we created a website that allowed users to select specific types of these layers as ‘blocks’ and drag them in a visual interface to create models, without having to learn programming and coding the math required for each of these layers.
As we were doing this, we heard about a startup competition hosted by NTU called ‘Ideasinc’. For early stage startups in Singapore, winners would receive $10,000 to further develop their ideas. Only 200 teams out of all applicants were accepted as applicants, and we decided to try our luck to be admitted to the first round. To our surprise, we were accepted as participants!
Looking at the other competitors, we figured that we would need to really up our game, so we decided to apply design thinking skills we had learnt about in IDEAS Hub to our product.
We showed our product to students and took feedback on what they liked and disliked about it. Eventually, through feedback and connections IDEAS Hub helped us create, we zoned in on our product as it is today. If I started now I would have done a lot of things differently! Some mistakes we made were being in love with our solution, without asking whether it would actually solve a problem, and not talking to users sooner.
Colin Peter, Technical Manager at the IDEAS Hub interviewed us for furtehr background on the MLBlocks Project:
What is MLBlocks?
MLBlocks is a platform to make machine learning simple for everyone. As a beginner in machine learning around a year ago, I realised that there were many barriers to entry for people just starting out, including needing to know how to code, needing to know complicated math, and the cost of powerful computers to train your AI.
To solve this, I teamed up with some friends to make MLBlocks. You need no knowledge of math or any powerful computers to train and use AI. All you have to do is create a dataset and upload your images, and MLBlocks trains and deploys model for you to use.
Currently, we’re still in a beta testing phase so that we can iterate into a great product once we launch, so we’d really appreciate getting beta testers to check it out!
How did you learn what you needed to learn to make MLBlocks?
I’d say it was this combination of self teaching and application of my knowledge that led me to creating it.
I started programming in my free time when I was 12, and started dabbling in creating projects just for fun in my free time. I mainly learnt by trying to create things that I found interesting and by using the internet to figure out what to do when I got stuck.
In the summer of Grade 9, seeing as how it was a rapidly rising field, I decided to look into how AI worked. After doing some courses online to learn the math, I began to create projects to implement this knowledge. Once I gained enough experience, I was lucky to get an internship in a startup here in Singapore to apply the knowledge I had gained. I now run an activity in school to teach students like me the basics of AI and programming.
How did you find the experience of winning ideasinc?
We never in our wildest dreams thought that we would win. We first applied on a whim; we didn’t even think that we would be accepted into the competition.
As the youngest team out of 200 selected teams of adults, it was very unlikely that we would even get through the first round. However, somehow we pulled through and reached the top 40.
Once we reached the top 40, we were called to NTU to pitch our idea in front of a panel of esteemed judges including professors from the University, after which we were accepted as part of the top 10 finalists.
The day of the finals arrived and we thought that our pitch went terribly. Compared to all the adults amongst us, we felt completely out of our league. As they announced the winners at the end of the day, we had our heads hung low, thinking that we had no chance after seeing all the other pitches. It was a complete shock, then, when we heard our team selected as “The Most Innovative Startup” in the competition!
I’m extremely grateful to have taken part in such an opportunity and to have had the opportunity to present something that I am proud of to such an esteemed group of people, but moreover, I really appreciate how much I’ve learnt about the journey.
What are some lessons that you've learned along the way?
Believe in yourself - cheesy but true. Who’d have thought that two teenagers could win a startup competition full of adults? As young people, we tend to assume that adults have all the answers. Perhaps, at times, we kids can too.
Persevere. Persevere. Persevere - We’ve been through three complete iterations of our product since we began. It’s the worst feeling in the world when you spend hours making something and no one uses it, but it’s important to push through and use your mistakes as a tool to improve.
Do what you find fun - Perhaps the most important thing in all of this is that I really enjoyed the entire experience. Unless I did, I don’t think I’d have had the perseverance necessary to continue when everything seemed bleak. So a takeaway would be to make sure that you do things you love, because those are the things that you’ll truly fight for.
Any advice for students wanting to create their own startup?
Firstly, make something simple that people will actually use. There’s no point creating some extremely technical product and telling everyone how amazing you are unless you can create something that actually solves a problem. What this means is, when creating a startup, focus first on making the thing before you market it and tell everyone how amazing it is.
Secondly, don’t get enamoured with your idea. Facebook started out as a website for college students to see who was in a relationship without asking them. Uber started out as a limousine service. AirBnB began as two broke dudes renting out their airbeds to people who couldn’t find hotel accommodation.
That leads me to my next point - iterate. Your initial idea will probably be way different that what it will end up being. Your job is to create stuff, test it with users, and improve based on feedback.
Finally, you don’t need to know how to code to create products! There are dozens of no code website builders out there, all you need is some ingenuity and a bit of hard work.