Abhisumat Kundu
Building intelligent systems that automate, predict, and scale human decision-making. Blending a developer's mindset with a researcher's curiosity.
About Me
My journey began with a simple curiosity about how technology shapes our world. That curiosity quickly evolved into a deep fascination with Machine Learning—the idea that systems can learn, adapt, and solve problems autonomously.
Today, as a 3rd-year CS student, I focus on bridging the gap between research and real-world application. Whether I'm developing predictive models, building GenAI tools like RAG systems, or creating full-stack applications, my goal is to build intelligent systems that scale human capability.
Beyond the code, I'm a builder who thrives on solving complex architectural challenges and optimizing algorithms to make a tangible impact.
My Experience
Research Intern
Indian Institute of Information Technology, Nagpur
- Worked on Multi Document Abstractive Summarization using Reinforcement Learning-Driven Transformer Model.
- Integrated T5-base with reinforcement learning to generate coherent abstractive summaries from multiple related documents.
Game Development Intern
OGames Studio
- Developed core gameplay mechanics for a mobile First-Person Shooter using Godot Engine and GDScript, contributing to 15+ functional game features.
- Self-taught Godot Engine and GDScript from scratch within 1 month, transitioning from zero experience to delivering production-ready gameplay features.
- Collaborated with an international team across 3 time zones.
My Projects
A novel approach to multi-document summarization using Reinforcement Learning. T5-base model trained on CNN/DailyMail dataset, achieving ROUGE scores of 39.5 (ROUGE-1), 17.8 (ROUGE-2), and 35.2 (ROUGE-L).
Open Source
Contributions to the community
Key Achievements
Best Use of Streamlit
MLH Hackathon
Won the official Streamlit category track by building an innovative data dashboard application (Ez-Viz).
numeth Package Success
PyPI / Open Source
Developed a Python numerical computations package that garnered 500+ organic downloads within 28 hours of release.
Skills
Education
MCKV Institute of Engineering
B.Tech in Computer Science & Engineering
Key Coursework
Beyond Code
District-level Swimmer
10+ Years experience
Chess Player
1350 Rating