Abhinav Rajeev Kumar

Abhinav

hiya, I'm currently a junior undergraduate student at SRM Institute of Science and Technology pursuing a bachelor's degree in computer science. my interests lie broadly in experimenting with model interpretability. more specifically, I am interested in the following topics:

trustworthy mL models via explainable ai
transformer architecture (like designing and researching on mixed model architecture/MoE/SBM architecture)
chain-of-thought prompting with GQA (Group Query Action)
agentic systems like compound ai

besides the nerdy stuff, i run a youTube channel. i have gone many sleepless nights watching football (Barcelona, Visca Barca), playing and listening to music (i don't have a specific music taste, i like all music but i do play the drums and occasionally the piano), and i also like watching anime and reading manga.

feel free to mail me at abku2504[at]gmail[dot]com if you want to chat.

RAO: Recursive Agent Orchestrator
Abhinav Rajeev Kumar*, Rahul Marban* (* = Equal Contribution)
github

ReportEase: A tool for generating insights from FIRs
Abhinav Rajeev Kumar, Varun Singh, Milind Mishra, Vidhi Singh
ppt| github

Shadow: A tool to analyze various crime data and provide visualizations
Abhinav Rajeev Kumar, Ahaan Sekhar Das, Varun Singh, Milind Mishra
website| github

Quantum Teleportation: Teleporting a qubit from one location to another
Abhinav Rajeev Kumar*, Vaishnav Manoj Kavitha*, Gayathri M (* = Equal Contribution)
pdf| github

Surge Protector: A drone built to detect stampedes via heat signatures
Anirudh Sathiya Narayanan, Max Weinreb, Abhinav Rajeev Kumar, Archna Sobti, Saatvik Billa
github

mar 2025:	i'm joining UIUC as a summer research intern working on problems with ai generated code under prof. Darko Marinov
oct 2024:	i become the syndicate of Satoshi Lab, leading the research on trustworthy ai and mentored 30+ members of the lab.
feb 2024:	i'm joining the quantum computing research lab as an undergrad student, working with prof. gayathri m
jan 2024:	conducted a research study with the quantum ai team to leverage insights of POVM-based information extraction to improve model interpretability
aug 2023:	started working with the chain-of-thought prompting to improve model understanding for multi-step reasoning tasks
may 2023:	i'm joining Google as a software engineer intern, working with the gemini team
feb 2023:	i join Next Tech Lab , a student run research lab at SRMIST, working on decentralized ai and quantum computing under Satoshi Lab.
apr 2022:	SuperImage achieves 100k+ downloads on Play Store
dec 2022:	i join SuperImage as a machine earning engineer Intern working on image upscaling using MNN DL Framework
sep 2022:	i'm joining SRM Institute of Science and Technology as a computer science undergraduate student

Understanding the Architecture of State Stream Transformers
Talked about Architecture of State Stream Transformers and how they use blending to improve context understanding.
video| paper| canvas

Understanding Chain-of-Draft Prompting
Talked about Chain-of-Draft Prompting and how it use excerpts or "drafts" to improve model reasoning.
video| paper| canvas

SRMIST

2022-2026

SuperImage

S2022

Google

S2023

NextTechLab

2023 - Present