What This Page Is For
Turns out I win a lot of things, work on a lot of things, intern quite a few times and take charge of numerous events. Also turns out adding all this info to an about page clutters it. Hence, this page contains the details of important projects I’ve worked on, achievements I’ve won, the internships I’ve done and positions I’ve held.
Covindia
From Mid-March 2020 to May 2020, I was the Project Maintainer for Covindia. Covindia was India’s first live district-wise tracker for Covid-19. I worked and led a team of 22 people comprising of 3 professors and 19 students round the clock to keep the website running. We got featured in 40+ news articles (online), 7+ news articles (printed) and I appeared on one interview on a live news channel on Television. We managed to hit 1 million views in 2 weeks after which it ran for 2 and a half months before shutting down.
I worked on setting up almost every working part of the site before shifting my focus over to the architecture and managing the Back End (website + infrastructure), all the while being the sole project maintainer. Contrary to previous projects and teams I had worked on / with, working on Covindia carried a substantial amount of social responsibility, took a lot of continuous & continual effort and dealt with real-time data.
Blog post explaining the journey in detail: Covindia
The last snapshot of Covindia: https://covindia.com/the-legacy
Smart India Hackathon - Winner
The Smart India Hackathon is a nation wide hackthon conducted yearly by The MHRD (The Indian Ministry of Human Resource Development) that brings in multiple state governments and organizations who offer problem statements that they face. 6 of us (plus 2 faculty mentors) worked incredibly hard from January to August to participate and win in the finals on August 1st - 3rd. We worked on a problem statement offered by The Government of Uttarakhand that entailed the prediciton of jobs in the future and set up a completely working proof-of-concept to present to the judges. We won Rs. 1,00,000 (~$1,350).
Blog post explaining our solution and work: rasp.team
Our solution and website: https://icecereal.github.io/rasp-redirect
HackTheCrisis - 10th Place out of 1,500 Teams
During 4th and 5th April 2020, HackTheCrisis - India was conducted by FICCI FLO Pune, Ministry of Electronics and IT, MeitY, MeitY Startup Hub (MSH) (Govt of India), Garage48, Accelerate Estonia, Robotex International (India Initiative), Science and Technology Park, Pune (Ministry of Science and Technology, Govt of India initiative) and APJ Abdul Kalam Centre. It was a nation wide 48 hour hackathon whose problem statement was to help solve problems created by COVID-19. There were 1,500+ teams that registered from all over India and the Top 300 got selected to participate in the hackathon. We participated and were selected to be in the Top 30, where we were judged in front of very important Government Officials. We won 10th Place!
The excerpt about this hackathon can be found at the bottom of this blog post: Covindia
ESummit Start-Up Sprint - 2nd Place
My friend and I participated in a 48-ish hour Start-Up Sprint hackathon conducted by The Entrepreneurship and Innovation Cell of our university. We made an Automatic Video Editor that personalized the primary edits of a sequence of images according to your previous editing style. It took note of every time you edited something new and learnt about your preferences using a bit of association learning. We won 2nd place and Rs. 15,000 (~$210).
The blog post that describes what we did: ESummit Start-Up Sprint
Dell EMC Hackathon - Winner
A 24 hour hackathon was conducted by Dell EMC in our University on 21st October, 2019. 3 of my friends and I took part in it taking up the problem statement titled AI-Enabled-Cart-Conversion. The gist of the problem statement is: “Customers add products to their online carts but sometimes don’t buy them. Some statistics show that if even 1% of the customers’ carts were converted from a abandonment to a success, major companies would make hundreds of millions of dollars. How would you use AI to make a customer buy the product”. We won first place and the four of us were offered a paid internship for summer 2020 for 2 months at Dell EMC.
This happened during a pre-blog-post-writing era and hence the only details that I can supply can be found on this page: ~/projects.
Undergraduate Research Symposium, 2020 - Rochester Institute of Technology
During my internship at Rochester Institute of Technology, I virtually participated in the URS 2020 held between July 30, 2020 to August 6, 2020 by making a video explaining the research work that I was doing under Professor Matthew Wright. At the time of making that video, my internship was still underway and hence I presented only the preliminary work.
The video I made can be found here: Evaluating the DynaFlow Defense with Website Fingerprinting - URS 2020 - RIT (Re-Upload)
Undergraduate Research Symposium, 2019 - Mahindra Ecole Centrale
I was selected to be one of the ten undergraduate students out of many that applied to present my summer internship work in front of a panel of three judges in the URS 2019 at Mahindra Ecole Centrale on 6th September, 2019. I presented my summer 2019 internship work titled: Usage of Genetic Algorithms for Scheduling Parallelizable and Non-Parallelizable Tasks.
Robotics Competition - Winner
Four of my friends and I took part in a two day robotics workshop and competition in the college fest of a nearby college somewhere in 2018. The competition was sponsored by Robokart.in and recognized by IIT-Bombay. It was pretty simple, to be honest. Simple usage of Arduinos to move a “car”. We decimated the competition.
Brave Browser Hackathon - Winner
Two of my friends and I took part in a 24 hour hackathon during our annual University’s college fest in 2017. The hackathon was sponsored by Brave Browser with the hackathon topic as blockchain, because Brave had introduced a new crypto currency called BAT and were trying to promote it. We created a decentralized music sharing platform called Harmonize that allowed content creators to directly sell music to customers, eliminating the middle men music distribution platforms. It’s probably note-worthy to mention that all three of us were freshmen competing against sophmores, juniors and seniors from our university and other universities. We won first place and ~$100, which is really less now that I think about it. We were just really glad to have won our first hackathon!
Internships
Summer 2020 - Cybersecurity Research Intern - Rochester Institute of Technology
The Cybersecurity Visiting Student Research Program (CyberVSR) is a 10-week summer internship program offered by Rochester Institute of Technology where undergraduates, master’s students and PhD students work with professors from RIT. I worked under Professor Matthew Wright and his PhD student, Nate Mathews. I continued to aid Dr. Wright and Nate even after the CyberVSR program was officially over until late December. During the 7 months of the internship, we worked on evaluating DynaFlow, a website fingerprinting defense proposed by researchers at MIT. We utilized numerous reduction mechanisms and statistically analyzed recorded traces to look for patterns. I made several parallel programs that deterministically analyzed these traces to eventually give us statisfactory results (which will not be written here until it is officially published). My programs saved us tens of hours of computation reducing a pipeline that would take 6 hours to test to a mere 40 minutes.
Links: RIT Blog and The CyberVSR page.
Summer 2019 - ML Intern - Matelabs
MateLabs is a Start Up in Bengaluru, India that has a platform called Mateverse, a Machine Learning Tool engineered for business professionals. It helps the customer design Machine Learning and Deep Learning models in a matter of minutes without touching a single line of code. At a point in MateVerse, a lot of plausible Machine Learning models are trained for a particular input dataset that the customer provides. My task was ensure that the training of all these plausible models took the least amount of time possible. A algorithm in MateVerse returned an approximate time that each model took and I used that as inputs to create 2 methods to split the training of these models amongst a given number of cores. Many Machine Learning algorithms are parallelizable (i.e. they can run on multiple cores) and many are not (i.e. they can run only on one). Accounting for all this, I built 2 methods to split the models amongst a varying amount of cores: A Greedy Algorithm and a Genetic Algorithm. The Genetic Algorithm proved to be the better of the two but it took far more time than the Greedy Algorithm. The Genetic Algorithm returned the best split that took the least amount of time and the least amount of cores. However, because the Greedy Algorithm is … erk, greedy, it returned the best split which contained the models running on all cores. Though my official role says ML Intern, I really focused more on the backend scheduling instead of ML. Despite this, my ML knoweledge came in very handy to build these schedulers.
This probably won’t make any sense to you, but eh, it’s a proud proof of my work:
-
Example Greedy Output: [6 cores, model 1, model 2, model 3], [1 core, model 0, model 4, model 5], [1 core, model 6, model 7]
- The 6 cores is meant for the parallelizable models. The Greedy Algorithm would have determined that this would have been the best split to run [models 0 to 7] on 8 cores
-
Example Genetic Output: [1 core, model 3, model 4, model 7], [1 core, model 1, model 2, model 6], [1 core, model 0, model 5]
- Genetic would have determined that this would taken the least amount of time (same as greedy) using only 3 cores. However, it takes more time to figure the best split.
-
Real Stats:
- 17 models
- Greedy: Time Taken To Execute: 0.5 seconds | Estimated Time To Train: 2,713 seconds | 8 Cores
- Genetic: Time Taken To Execute: 7.8 seconds | Estimated Time To Train: 2,713 seconds | 3 Cores
- Generations: 50
- Initial Population: 8
- Chance of Mutation of Children: 40%
- Chance of Mutation of Genes: 50%
- Number of Best Children chosen: 4 from Crossover, 4 from mutation
Matelabs’ website: https://matelabs.ai
Summer 2018 - Summer Intern - GlobalEdge Soft
GlobalEdge Software is a software development company who focus on the Intelligence of Things. I worked at their Bengaluru office in summer 2018 and worked on three tasks over the period of 2 months.
- Gesture Recognition using Deep Learning
- This was my first attempt ever to get into the AI/ML domain and it took me around a week to get the hang of it. I understood the need of Machine Learning and Deep Learning and vaguely understood the math that drove it across various models. The task was to implement a Deep Learning program that could identify gestures via a camera. I used a VGG16 model over a dataset I created and reached 60% accuracy, which I think is quite good considering it was my first time into the domain. Shows also how easy it is to get into it.
- Stereo Vision using Two Cameras
- Using OpenCV and two cameras, I was able to implement a program that calculated the distance of an object from the mid-point of the two cameras.
- GUI for a Software Program
- I wrote a GUI for a software program in ~4 days that triggered processes to test a few programs created by some people. Afterwards, it displayed a report that contained a comprehensive analysis. I learnt and used PyQt5 to make it.
GlobalEdge Soft’s Website: https://www.globaledgesoft.com/
Positions Held
- Representative of Media and Technical Affairs, Student Council - Mahindra Ecole Centrale 2019 - 2020
- Part of the Student Council. Can’t get any bigger than that.
- Co-Organizer - TEDxMahindraEcoleCentrale - 2020 & 2021.
- I was chosen to be The Co-Organizer, the highest role in TEDxMahindraEcoleCentrale alongside The Licensee. We were a 30 member team comprised of freshmen, sophmores and juniors and worked for 6 months on the event’s logisitics, finances, sponorships, promotions + designs and the handling of speakers. It was set to happen in April 2020, but didn’t happen because our country went into a lockdown due to the Covid-19 pandemic a week or two before the event. However, we conducted it in 2021 on YouTube and I was the technical coordinator of the event: TEDxMahindraUniversity 2021 - Alt + F4
- Secretary - Enigma, The Computer Science Club - 2019
- I was chosen to be Enigma’s Secretary as a sophmore and as a junior. Enigma is probably the club that I put most of my effort in and probably the club that I love the most. We started out with a few members (less than 10) and currently have a much higher number of members. I took sessions on the basics of C, an introduction to Python, the usage of the Arduino Uno, interfacing with a Raspberry Pi, the math behind multiple Machine Learning algorithms and so much more that I really can’t remember. We also organized massive hackathons that had an organizing committee of 5 people but a participation of 250 students. We conducted “Hackweeks”, week long hackathons with the topics given by our University’s professors for students who are looking to enter a new domain. Fun Fact: 2 of us had to get back 250 pizza boxes in 17 massive bags in one Uber Go for the a hackathon. It was … very uncomfortable, yet very memorable and it smelled really good.
- Head of Logisitics - TEDxMahindraEcoleCentrale - 2019
- Prior to being chosen as the Co-Organizer for TEDxMahindraEcoleCentrale 2020, I was the head of the logisitics team.
- Organizer of tons of college events:
- Point-Blanc was one of the first large gaming events in our University. I’m glad to say that my friends and I were the organizers of it.
- Freshers Night. At this point, these are just really minor things that no one really cares about. If you’ve made it this far, I’m in awe. Anywhooo… I was the organizer of the Freshers Night for the 2018 batch.
- There are more, but those are so small that it’d be silly to even mention them.
- Junior Editor - MECRise - Annual College Magazine - 2018
- I was one of the junior editors of MECRise, Mahindra Ecole Centrale’s Annual College Magazine. I became an active contributor the next year.
- Team Lead - Economics - 2017
- This may seem really small and super unimportant; but for me, this was a semester long effort which we managed to come out on top of and thus, I am including it here whether you like it or not. Time: Semester 1. Class: Economics101. Professor: Dr. Praveen Malla. We were going to have a semester long Mock United Nations that was slightly modified to accomodate the course. I was chosen to be the Team Leader and for some reason, we decided to choose Somalia as the country we wanted to represent, giving ourselves handicap compared to all the other teams who chose developed nations. Long story short: we annhilated the competition, came first overall, got the highest points for 3 out of 4 presentations and I got a perfect 10.