SIH - My stepping stone into the world of hackathons

My team and I won “Smart India Hackathon Software edition 2022”, and we honestly haven't left a single chance to rub it in everyone's face. The reason is simple to say the least. We are excited. Sih was our first hackathon and not only did we go into the world's biggest hackathon with no prior experience, we bagged the first prize.

“It's not the destination, it's the journey”

Our journey began way back in January. One fine day, Janvi casually approached me to tell me she had given my name for a hackathon team. I felt a flurry of emotions. I was excited, scared and nervous. I met my team leader Dinesh and other team members Vinit and Prince shortly after. Thus, the team was formed even before we knew the problem statements. However, we were still short of one team member. After a lot of searching, we found a 3rd year student(Unnamed) to be a part of our team. We worked relentlessly, upskilling ourselves, discussing strategies, and working on various aspects of a hackathon. Our schedules included G-meets till 3 am after a full day at college. And not once did fatigue overpower our passion.

When the problem statements began to roll out, we started out with a problem statement by the ministry of Rajasthan, working on a queue management system. A few days after working on this statement, we had our research, our solution and a prototype of our app. And then things slowly took a turn. We began hitting dead ends with our project. It was infuriating, exhausting even.
Then our college released the officially selected problem statements for us. We were back to square one. Among the 20 or so problem statements provided to us we were all drawn to one. Problem id: “NR1174”. Our problem statement required us to build smart legal aid, to reduce the pendency of cases, increase the efficiency of courts and educate the masses about the working of the legal sector.

Our first step was to research: We read research papers, news articles, e-court manuals, all in a frantic attempt to understand the underlying problem. Our second step in research was to talk to people. We set out in and out of college, on streets, buses and every nook and corner of our neighbourhood, interviewing people about their experience with courts, problems they faced and solutions they would suggest. It was in one of these interviews that we met a young man, who told us, “Doosre deshon mein courts ache hai, par ye India hai na” (other countries have a better court system, but this is India). The validity of that statement is of rather little importance here, because that statement sparked something in us. It was at that point that this problem statement stopped being just a “hackathon thing” and began to mean so much more to us. That day, my teammate, Vinit said “Doosre deshon ke paas hai toh kya, hum bhi bana lenge!”. (so, what if other countries have it, we will also make it.)

With that began the real journey of team hum bhi bana lenge. We immersed ourselves in our work. Contacting judges, advocates, law students to reach to the root cause of our problem (this will be important later on.) There was a point where we had emailed over 100 judges and contacted over 15 advocates. [I’d like to take this space to thank Saumitra Singh Bhataudia (probation judge at the Ahmedabad high court), Mohammed Ali (criminal lawyer in high court of Karnataka), Mehul Thakkar (registrar at Baroda district court), Vaibhavi Venugopal and Rahul Ranjan (2nd year law students)]. Without the help of these law-versed selfless individuals and their insights in the litigation system, it would have been impossible for us to garner the information that we did and thus build our project.

Slowly we inched towards march, our research was done. We knew the product as well as the back of our hand. We cleared the internal hackathon organised by our seniors at point blank, submitted the ppt to SIH portal, and with that, we were done. We decided to take a break from SIH while awaiting results and work for our upcoming college exams.

Results began rolling in around the first week of July. We were all on the edges of our seats. The first set of results rolled in, and our problem statement was nowhere to be seen. The second week rolled by, our nails were chewed with anxiety but yet again, the results for our problem statements had not been announced. We were scared, perhaps none of the solutions had been selected for our problem statement? It wasn’t long before the third set of results rolled in, and this time we were jumping in our rooms. We had made it, we had gotten selected. Our excitement was through the roof. Out of the 20 teams that had applied for our problem statement, we had emerged as the top 3.

Soon after the results were declared we got the tentative dates for the hackathon and learnt that one of our team members would not be able to make it there. We felt dejected. Did this mean we wouldn’t be able to go? It was no use whining so we set to action. A new hunt for a new member. On hopped Dheeraj, our final member, our ML expert.

This would be a good point to begin explaining our initial approach to the problem. We were sure that we didn't want a predictive algorithm for three main reasons, 1. Any predictive algorithm would learn from previous data, making it biassed owing to prejudices from the past. 2. We wanted our product to help judges make faster decisions, but not at the risk of compromising on justice. FAIR AND FAST, was our motto. 3. Outright having AI give judgements was a step that would take a lot of time for both the common man and jury to trust and appreciate, and we did not want our product to give unsatisfactory results like the digitise courts movement from 2013.

The solution we came up with was intricate and promised results. It involved a scheduling system, a mobile app and a web-app.
Our notification system (built with Twilio for the sake of the hackathon) would send notifications a day prior to the hearing date, to all concerned authorities and relevant parties, and use a confirmation system (yes /no reply system). Details of the Parties who confirm their presence will be sent to part 2 of the solution. Parties that deny their presence, will be rescheduled to a different date. Each party gets up to 3 turns to postpone their hearing owing to emergencies, after which they are mandated to attend.
Part 2 of the solution takes in confirmed party details as input and generates an ai score based on a multitude of parameters like case type, case stage, and case age. This score generated is used to schedule the Judges day. This scheduling is reflected in the judges mobile app. The judge can see any case details relating to a particular case, these details include the entire case affidavit, the summary of the case up until that point of jurisdiction and a list of precedents (cases of similar nature that have been resolved by any court higher than the court seeking it) with a similarity score.
A web app questionnaire for the common man, where within a few clicks, users answers questions relating to their cases, a report is generated at the end, providing valuable information like average timeline of case, average cost, laws applicable, documents required, a YouTube link for easy understanding etc.

With the game plan ready, the grind that would lead us to the winning moment began. With our persistent requests we were able to secure a room to work in. Work hours ran from 9 am to 7 pm, at which point department doors were shut and the only way to leave was with us jumping over walls. We faced a multitude of challenges during this entire process, mainly due to the lack of technical knowledge. Our team wasn’t the strongest technically, but we would not that deter our confidence. Vinit and Janvi worked on the mobile application. Dheeraj worked on the ml models, while I worked on deploying them and the product pitch. Dinesh worked on the design and website aspect of our app, while Prince worked on building the dataset, making the product more accessible and occasionally on the front end of the website. We struggled the most with APIs and Deploying the ml models. The model gave multiple errors when deploying on Heroku and when it finally did deploy, it was too large to deploy on the free version. The API from backend, was made with tokens and empty functions, proving to be a very hard task to use in the android app.

With barely any of the product ready, and new challenges popping up every step of the day, it was time for the finals. I mean, final internals exams. Right before SIH. But we managed to waddle through internals, SIH prep and made our way into the humongous campus of ACS college of engineering. Once the hackathon began, everything seemed to pass by in a blink. We were the last problem statement, seated between our competitors. Seeing both teams have a mentor, instilled some fear in us, but not for long. Fear was quickly replaced by Hunger, Drive, Passion. Our first review went extremely well, we were applauded by both members of the jury, and were feeling pretty confident. The second review was the exact opposite. We fumbled, our project had crashed and we couldn't seem to figure out the problem, and we couldn’t answer the questions asked by the jury. It was disheartening. Few of us were already in tears. This is where our team leader stepped in to the rescue, he told us we needed to calm down. We went outside, letting the cool air soothe us. We contacted seniors and friends, seeking help. It was then that one of the jury met us, told us we were on the right track and asked us to keep moving forward (and asked to take a selfie with us). That changed everything. We were back in form. Better actually. We were determined to make this work. In the final evaluation, our project was still not fully functional. We had bits and pieces of code breaking. Errors. Empty Screens. Blank Faces.
(Remember I told you, the research would be important later on)

With what we had, we were able to convince the judges and show them the utility our project had. All the research we had done began to pay off, and despite not having a fully functional model ready, judges saw and appreciated the groundwork we had done. We were not building a product for a hackathon, we were building it to solve a major problem in our country, and that earned us the brownie points that we needed to win.

Within an hour after our final presentation, we had reached the final deciding moment. We waited with bated breath as the results were being announced. “The winner for ps id NR1174, our last problem statement is…”, we were on the edge of our seats. Sweat trickling down our backs, “Team hum bhi bana lenge”. There was an uproar, emerging from us of course. In no time we were sprinting towards the stage, so fast Usain Bolt would’ve been scared and then we were there. On stage receiving our cheque, taking blessings from our jury and other officials. Out of the 36 hours allocated to us, we had been up for 32+ hours. Despite that, as we held our prize, the expected fatigue never set in. We were smiling end to end, dialling phone numbers of parents, friends, jumping with excitement. We had done it. We had won our first ever hackathon!
In the span of six months, I had won a hackathon, made friends for life, and I was only just getting started!