Hi 👋 :)
I'm Rashmil Panchani. I am an athlete at heart, with an entrepreneurial spirit and always up for learning new things. Being able to adjust to situations makes me a very flexible asset to anyone who hires me and to my collegues. My passion lies in creating and developing new things which includes WebApps, Embedded Systems and applying Machine/Deep learning models in various projects. My research interest lies in Scene Perception and Cognitive tasks of Natural Language Processing. Apart from a techie, I am also a photographer(Long-exposures being my favourite).
August 2021 - December 2022
University of Illinois Urbana-Champaign
GPA: 3.94 / 4
Master of Computer Science, graduated December 2022 with a GPA of 3.94 / 4.
August 2017 - May 2021
DWARKADAS J. SANGHVI COLLEGE OF ENGINEERING
CGPA: 9.963 / 10
Bachelor’s in Computer Engineering with a CGPA of 9.963 / 10.
2021
My final year BE project. This project uses IoT to smartly take care of the plants as well as provide Information on the plants who wants to know about them. Main features include Realtime sensor updates, Live Video feed of the garden, Plant Disease detection through OpenCV. It is a React based web app that leverages the power of cloud computing to keep the production code managable.
2021
This was the end project for the HSE NLP course on Coursera. I was tasked to build a chatbot which was able to understand the users input and classify the intent and then based on the question, give appropriate responses. The bot was made to be run on Telegram and hosted on an online instance such as AWS. Currently it is not hosted anywhere ans is available for use and training on Github.
2019
This project was built in the qualification round fron the college, so that we can send our proposals to the SIH committee. The project goal was to create system for online lectures that will help the teacher in improving their lectures and provide the students with more engaging content. The project leverages the power of high speed of WebRTC and AioRTC in Python to provide real time analysis for the realtime streaming and processing the video feed. While processing the video feed we analyze the video and audio to extract features such as how long was the attention span of the student and when the student was the most attentive and the least attentive. For the Audio part, we provided detailed analysis of what percentage of the online lecture was one way (Teacher to Student) or two way (both Teacher to Student and Student to Teacher).
2019
The tower of age of warring empire game has 10 levels and spawns a random map every time, I found this very boring as I had to complete these levels nearly 50 times to level up my heroes from 1-120-ish level and the max level of heroes were 400, that means 50\*10 = 500 times. My solution was this, using OpenCV and template matching, I designed bot that on given certain restrictions, it will utilize gold coins or energy cards accordingly and automate the attacking process. This reduced my efforts as I don't have to play and my heroes get levelled up to 400 in a matter of few hours. Another use case that I developed is the Hydra bot, a Hydra event in the game has 1000 levels, each having 10 sub levels thus, resulting in 10000 attacks, I applied some similar strategies but with deadlock conditions and now I can clear approximately 800 levels in a matter of 6 hours.
2019
The main goal of the project was to develop a model that was able to detect various signs used by the mute people to communicate with people who do not know how to interpret it. Hence I thought of developing a model that can be integrated in the mobile app and can help the user to decipher what the other person is trying to say without the help of a translator. The end goal was to integrate the ability to use optical flow analysis and be able to convert the signs into Natural Language and give an audio output but due to the lack of computation resources we could not complete the entire project.
2018
I made an emotion recognizer that scans the frontal face using Haar Cascade and then uses the faces detected to look for what emotion class does it belong to. The detection is fast enough to be run at ~30fps.
2018
e - Yantra Robotics competition: I was the team leader and we built a line following, color sensing bot with picking and releasing mechanism using Atmega2560 which traversed the arena using Djikstra's algorithm. For this competition, we would be given some boxes with the details of where they should be placed by the robot and our goal was that the robot should follow the black lines and pick and place the boxes in their desired locations. Along, with completing the task in minimum time, the robot should also avoid any collisions with the boxes and the obstacles that may be placed on certain routes. We ended up being the runner ups of the competition.
2018
This project was aimed at creating my own personal website where I get to showcase my skills and projects.
August 2022 - Present
National Center for Supercomputing Applications (NCSA)
Maintaining and building the pyincore python package. IN-CORE is a Hazard and Disaster Management planning tool that helps users plan their management strategies by providing powerful visualization tools and datasets. The main goal of IN-CORE is to manage and mitigate the community from Natural Hazards and Disasters by creating management policies that are fueled by state-of-the-art algorithms designed to strengthen the Resilience curve.
August 2021 - August 2022
National Center for Supercomputing Applications (NCSA)
Worked as a researcher and full-stack web developer on a visualization and analytics tool for studying Oceanography and Marine biology — helping researchers visualize species distribution in the ocean and study the changes that have happened from 1873 to present.
January 2022 - May 2022
Department of Computer Science, UIUC
Formulated algorithms to model data in Python to generate multi-dimensional graphs containing 200,000 nodes. Devised an end-to-end system that visualizes bibliographic networks and helps conference/journal panelists infer possible conflicts between paper reviewers — enabling informed decisions in reviewer allocation.
May 2019 - July 2019
Phionike Solutions: Design-Tech Studio
My work at Phionike at first involved laying down the foundation of Research & Development of AI department in the company. During my time there I did a research in the VUI(Voice user interface) sector and what impact does it have on the various sections of the society. I basically researched about how a VUI system works and to do that I studied how the famous Amazon’s Alexa works. I also built a skill which fetches blogs and reads them out for you. Other than VUI systems, there were several research ideas that the company had in store to work on, my job on them were to create proper documentations for future references.