chinmay kamerkar

The root cause of my choice to pursue my undergraduate degree in Computer Science is in similitude with the nature of my surroundings: the inspiration for my intuition and logic. I have had an innate fascination towards understanding human psychology and intrigue towards algorithms. The remarkable change that has been brought about by the technological advances in the field of Computer Science has motivated me to delve deeper into the field.
I applied the concepts I learned in my undergrad curriculum to explore the applicational ambit of my field. In the Smart India Hackathon 2019, we developed a portal to track the availability of officers in AICTE India which was an immense learning experience. Its basic human tendency to be more comfortable posting queries than navigating for the answer. In accordance with this, I designed an intent-classifier for relevant search on the portal. We built and redesigned our PWA(Progressive Web App) within just 30 hours. A decent understanding of Object Detection and curiosity towards Natural Language Processing engendered the idea for my final year project. I designed a people analytics platform that enables managers to improve performance through targeted use cases for optimizing crowd flow in closed built environments and generate contemporary insights like forecasts and heatmaps. For object detection, I used the MobileNet v2 architecture which uses an inverted residual neural network(NN) with a linear bottleneck which is an embedded device focused NN architecture. We saved 9 times more computational power at the cost of just 8% accuracy. This project taught me a great deal about analyzing the trade-offs in a certain solution based on the context of the problem.
To utilize the opportunities in industrial settings, I worked as an intern on interesting projects. I was a project trainee at National Informatics Center (NIC), wherein I was assigned a Proof-of-Concept (POC) project to develop a Blockchain for stakeholders involved in Export activities. We developed a shared distributed ledger using Hyperledger Fabric which was economical, immutable, decentralized and established non-repudiation to overcome shortcomings of the traditional EDI message exchanges. At Course5 Intelligence, I developed a Sales-Intelligence tool to integrate information from cold-calls, SalesForce and emails into a single central dashboard. I developed a sentimental analysis module using the LDA classification algorithm to be used in the next phases of the project. The project required me to learn Flask(a micro web-framework in python) and follow PEP 8 guidelines. It taught me how to ensure the efficacy as well as the reusability of the code. Consequently, I was also delegated to contributing to and analyzing code efficiency in the python backend modules of an on-going project.
These experiences and challenges have helped me improve myself on technical grounds and also as an individual.