Vaibhav Balloli

Vaibhav Balloli

Ph.D. student

University of Michigan

👋 Hi, I’m Vaibhav Balloli, a Ph.D. student at Realaize Lab with Prof. Elizabeth Bondi-Kelly at the University of Michigan, Ann Arbor. I’m broadly interested in the intersection of sequential decision making and multi-agent systems with applications that focus on benefitting the society. Previously, I was a Research Fellow at Microsoft Research India, I’ve worked on HAMS, Project Vasudha and Project VeLLM. I’m a proponent of Open Source and also pretty active in developing in and contributing to Open Source, so do checkout my work on my GitHub.

Interests
  • Multi Agent Systems
  • Sequential Decision Making
  • AI for Social Good
Education
  • Ph.D. student, 2023

    University of Michigan

  • B.E. in Electronics and Communication Engineering, 2020

    BITS Pilani, Hyderabad Campus

Recent News

All news»

  • May 2024: One paper accepted at IJCAI'24 Human-Centered AI Track (<5% Acceptance Rate)!

  • April 2024: One paper accepted at CV4Animals Workshop at CVPR 2024!

  • April 2024: Second place in Google x MHacks Hackathon (Team: Me, Prize: $1500) for building the next generation package management in the age of Generative AI.

  • September 2023: One paper accepted at NeurIPS 2023!

  • August 2023: Joined UMich, as a Ph.D. student!

Experience

 
 
 
 
 
University of Michigan
Ph.D. Student
University of Michigan
August 2023 – Present Ann Arbor, MI
Graduate student at the CSE Department at Realaize Lab.
 
 
 
 
 
Microsoft Research
Research Fellow
Microsoft Research
June 2021 – June 2023 Bengaluru

Projects:

  • HAMS - Automated License Testing (AI for Social Good)
  • Vasudha - EnCortex package that provides optimization and decision making for improving sustainability of energy producers. (Currently integrated as a product at Energy & Mobility, Microsoft)
  • VeLLM - Improving multilingual capabilities of Black-box LLMs.

Recent Publications

Quickly discover relevant content by filtering publications.
(2023). Learning Runtime Decisions for Adaptive Real-Time Perception. NeurIPS.

PDF Cite Code Project DOI

(2023). Breaking Language Barriers with a LEAP: Learning Strategies for Polyglot LLMs. In Review.

PDF Cite Project DOI

(2019). Video Streaming using Scalable Video Coding over Opportunistic Networks. In WiSPNET ‘19.

PDF Cite Code DOI

Recent & Upcoming Talks

Decision Making for Social Good applications
Presented my research as a Research Fellow for 2 years at MSR

Projects

*
HAMS

HAMS

Automated License Testing framework

NFNets and Adaptive Gradient Clipping

NFNets and Adaptive Gradient Clipping

Implementation of the DeepMind paper in PyTorch with the models and a pytorch optimizer wrapper for AGC.

Offlax

Offlax

A library implementing popular Offline Reinforcement Learning algorithms in JAX

Ripple

Ripple

PyTorch library for Risk-aware ML

Vasudha

Vasudha

Optimization Framework for New-generation renewable energy producers

VeLLM

VeLLM

Project VeLLM at MSR India aims to empower everyone, no matter what language they speak, with the transformative power of Generative AI

Contact

You can always reach me through email or twitter.