My C4GT Experience
Update for all:- Ik this is a long read, but I got selected so it might be helpful :) I have been selected for the Ministry of Electronics and Information Technology project “LLMs for Question Answering” .
Hi everyone, today I’ll be sharing my experience of writing two proposals for iGOT Karmayogi organization under C4GT program. If you don’t know what is C4GT, think of it as being similar to GSOC but with a focus on Digital Public Goods. You can find out about it here :- https://unstop.com/competitions/code-for-govtech-2023-samagra-governance-675801.
DPGs have always fascinated me as to how our 12th-grade mark sheets can be accessed quickly on DigiLocker, how easy it is to travel via airports because of DigiYatra, and many more such examples, so this program was just the right opportunity for me to explore such impactful projects.
Finding a Project:-
Finding the right project was no easy task. I skimmed through the projects based on my preferred tech stack in Python, Natural Language Processing, Flask, and Streamlit I came across 3 projects that well fit my skillset– LLM question answering by iGoT Karmayogi, cQube Chat, and Health Exchange Python SDK Project. I also looked around AI Tools- Neural Coref and Document Uploader.
This step turned out more of a lesson for me and that was not to self-reject and give it a try. When I started exploring the projects, I couldn’t help but see the overwhelming number of talented people expressing their interest. These individuals had impressive credentials and experience in the exact technology stack needed for the projects and it made me feel very nervous about applying. Joining the discord server, I could see the conversations between other contributors and mentors that happened about 2 weeks before I even started exploring the program, let alone started working on it. I started feeling that my selection rates were now very slim. Giving up on this open-source program and dedicating time to DSA seemed a more efficient approach.
In a moment of reflection, I realized that such reputed open-source programs aren’t just about getting selected. They’re for learning, creating, and collaborating with other talented folks and this program is all about writing code for DPGs, what’s better than contributing to the tech sector of our own country? Most people give up at this stage, I was going to as well, so future contributors don’t make this mistake. Just the proposal-making stage of this program has taught me a lot. Also, Lord Krishna has taught us that all we can do is devote ourselves to the task at hand and do the hard work honestly, the result is what he decides for us so why worry about it? so with this mindset, I moved forward.
Proposal Writing:-
I explored all three of the projects and found that they didn’t have an existing codebase and had to be developed from scratch. While I had previously contributed to large open-source scientific projects like NetworkX, I realized that developing project plans for population-scale was a whole different ball game. It soon hit me that these three are very different and enormous projects, and I won’t be able to develop proposals for all of them. I re-read them again and did my research, I tried making project flow charts to draw out my initial thought process and ideas on excalidraw much like:-https://excalidraw.com/#room=9b53acab57272d5330f8,4oOurEW3l5ppeiGTt_YY4A for the projects and finally landed on iGoT Karmayogi’s LLM’s for Question Answering Part I and II as my final project. I figured it’s better to send one detailed proposal for this project and do it justice than send out average proposals for all three projects.
I felt like I had the best understanding and ideas for this project. This project is divided into two parts, so I made two proposals one for each part. We had been given a sample proposal and I followed that template. I referred to some “How to make proposals for GSOC” youtube videos too for this. The mentors helped by giving us pointers on how to work on the proposal. iGoT Karmayogi’s mentors were very helpful and conducted Microsoft Teams sessions for us to ask our queries and guided us about the project details.
Overall Thoughts:-
Developing a population-scale project plan was a new learning experience. Just deciding the tech stack to use for this project meant taking into consideration the scaling, costing, and infrastructure requirements for the project. These were things I didn’t pay much attention to before, but now they were really important. Since this project is starting from scratch, it is crucial to test everything thoroughly and make good documentation. These were steps that I often overlooked in smaller projects, but now I realized how essential they were. It truly taught me how to create a good software engineering project and look at the bigger picture.
Overall, just through the process of crafting these proposals I read many new research papers, blogs, medium articles, Kaggle notebooks, etc to create a good solution and that has only made me better at NLP. If anything, I have more confidence in my skills than before this program.
After spending so much time researching and creating the proposal, I still felt that it missed many important pointers so when I heard that the deadline got extended by 12 hours, I was happy and made several changes to the proposal and submitted it again. Honestly, I have read the proposal some 20 more times again after the submission and each time I find new mistakes or some new additions like Batch Processing could’ve been made, etc etc.
As I await the results, my mind continues to wrestle with anticipation and uncertainty. Yet, in the midst of this internal struggle, I am overwhelmed with gratitude for being a part of this program. I am immensely thankful for the chance to create proposals for such impactful projects and for the invaluable skills I have acquired along the way.
No matter what the outcome may be, this experience has shaped me in profound ways. It has taught me perseverance, resilience, and the importance of continuous improvement. I am grateful for the opportunity to contribute to the advancement of technology in service of our country, and I eagerly look forward to the next steps in my journey.
