about

Hi, I'm Ashutosh Sathe, I'm a first year graduate student at IIT Bombay. I did my undergrad studies in computer science from College of Engineering, Pune. Prior to coming to IIT Bombay, I worked as an analyst at Aspect Ratio where I explored temporal machine learning models for COVID-19 forecasting. I'm interested in building interpretable and robust machine learning models that are ready for real-world applications

projects

1. scarpet-nn - A set of tools, utilities and libraries to run neural networks in Minecraft. Because Minecraft computers (with/without redstone) can be very compute intensive, this project focusses specifically on binarized neural networks which are super light on compute. You can also convert models trained in PyTorch to a schematic for easy rebuilding in Minecraft. Read more about the project in this blog or watch the demo video on youtube
2. yacv -- Yet Another Compiler Visualizer - A manim and graphviz powered parser visualizer. Given a context-free grammar, the input string and the parsing algorithm (LL(1), LR(0), SLR(1), LR(1), LALR(1)), yacv can produce a full step-by-step parsing animation video, LR(1) automaton, parse tree (as an image) and output the entire parsing table as a CSV file. Read about the project (and some parsing trivia) in this blog or watch the demo video on youtube
3. spreaddit - "spreadsheet + edit = spreaddit" as the name suggests is a very simple, open source, lightweight spreadsheet editor completely written in C. This was my course project for Data Structures and Algorithms course in my undergrad. The project uses GTK+3 for windowing but can also run headless and has following functions:
   1. Read/write CSV/ODS files
   2. Sort a file on a specified column
   3. Date format manipulation (identify and convert dates from DD/MM/YYYY ↔ MM/DD/YYYY etc.)
   4. Arithmetic expression evaluation (e.g. C3 = C1 + C2) based on cell data

hobbies

education

register