Jul 2026 – Present
Internship
Approximate LabsWorking on reinforcement-learning research.
About me

I started by implementing ideas from scratch and kept following the questions that felt hard. That path led from foundational ML systems to multilingual NLP research, open-source tools, and products built around learning by doing.
B.Tech, Indian Institute of Engineering Science and Technology, Shibpur
2023 – Present · CGPA 8.88
Python, C, C++, PyTorch, TensorFlow, NumPy, Pandas, Git, Linux
Machine learning, deep learning, multilingual NLP, LLMs, research systems
Research, systems, and the work in between.
Jul 2026 – Present
Working on reinforcement-learning research.
May 2026 – Jul 2026
Designed and developed the testing infrastructure for mock validation of router software, establishing the framework from the ground up. Automated internal engineering workflows, significantly reducing manual intervention and accelerating routine development tasks.
Jan 2025 – Apr 2026
Worked with Dr. Raj Dabre on multilingual NLP, transliteration, low-resource language modeling, and multiple accepted papers.
Dec 2024 – Jan 2025
Worked on machine-learning and deep-learning research problems.
Jul 2024 – Sep 2024
Worked on foundational ML systems and experimentation.
What I value while working on difficult problems
I like following difficult questions far enough to turn them into experiments, code, and publishable results.
I care about careful evaluation, reproducible systems, and the details that separate a plausible result from a dependable one.
I learn by shipping and sharing implementations, tools, datasets, and explanations so the work can be useful beyond a paper.
Reviewing, explaining, and sharing the work.
Lossfunk: Why LLMs Think in English and How It Affects Their PerformanceLink coming soon