Vaibhav Balloli

Applied AI Researcher · Computer Science Ph.D. Candidate, University of Michigan

👋 Hi, I'm an applied AI researcher and a Computer Science Ph.D. candidate at the University of Michigan. I design and deploy interactive AI systems that help solve specific societal challenges across healthcare, wildlife conservation, road safety, and education. My research focuses on topics around LLMs, Reinforcement Learning, Representation Learning, and human-AI interaction.

Previously, I have worked at Microsoft Research: once as a summer intern working on improving interactivity of LLM agents and then as a Research Fellow for two years working on scaling HAMS, Vasudha, and improving multilingual capabilities of LLMs at VeLLM.

Always glad to hear from people working on similar problems: research collaborations, student questions, or just a good argument about evaluation. Reach me at balloli.vb@gmail.com.

📄 CV Email GitHub Google Scholar LinkedIn Twitter Friends

News #

Apr 20, 2026
One paper accepted to KDD 2026 in the Datasets and Benchmarks track.
Apr 20, 2026
One paper accepted to CV4Animals @ CVPR 2026, in collaboration with BeeryLab at MIT.
Apr 15, 2026
One paper accepted to FAccT 2026.
Apr 1, 2026
One paper accepted to Interactive Health 2026.
May 15, 2025
Started my internship at AI Frontiers, Microsoft Research.

Systems & deployments #

The same work as the publications below, organised by what was built rather than where it appeared: what I contributed, where each system ended up, and how far along it is.

HAMS: automated driver licence testing
Deployed Microsoft Research India 2021 – 2022

Societal Impact through Cloud and AI Fellow. Worked on the computer vision, bandit, and localisation components as the system scaled.

Smartphone-based automated licence testing deployed across India, serving 300,000+ candidates and replacing a subjective manual assessment with an auditable one.

Computer vision On-device inference Bandits Localisation

Project Deployment Press
PATHFinder: prenatal care planning agent
In progress University of Michigan 2025 – 2026

Lead author. Designed and built the four-stage agent: patient intake, dynamic interaction, plan synthesis, and clinician oversight.

Conversational agent that produces individualised prenatal care plans aligned to ACOG's PATH guidelines and surfaces local resources. Frontier models scored against expert rubrics across five clinical dimensions, with clinician evaluation in Michigan health systems underway.

LLM agents Long-context optimisation Clinical rubrics

Project arXiv Paper →
CHAIR & SEEK-CBM: steerable wildlife re-identification
In progress University of Michigan · MIT 2023 – 2026

Lead author on CHAIR; co-author on the SEEK-CBM follow-up with the Beery Lab at MIT.

Concept-bottleneck retrieval that lets conservation experts intervene mid-search, built for platforms like ElephantBook and evaluated on elephant populations in the Masai Mara. SEEK-CBM's modern backbones improved retrieval performance by 497% over CHAIR.

Concept bottleneck models Self-supervised vision Human-in-the-loop retrieval

Project Code Talk Paper →
VeLLM: multilingual LLMs for the global majority
Prototyping Microsoft Research India 2022 – 2023

Research Fellow. Co-led the dynamic prompt/model/embedding selection work that became the COLING 2025 paper.

Platform work behind LLM applications piloted in schools across India, improving coverage for non-Latin-script and low-resource languages without additional training.

LLMs Multilingual evaluation Prompt & model routing

Project Context Paper →

Research #

I build interactive AI systems that people can steer: agents that ask before they advise, retrieval a domain expert can correct mid-search, and models that reconfigure themselves at run time. The hard part is rarely the model. It is deciding what a system should be uncertain about, what it should hand back to a person, and how anyone would know when it is wrong. That question runs through my work in healthcare, wildlife conservation, road safety, and education, and sits at the intersection of large language models, reinforcement learning, representation learning, and human-AI interaction:

Interactive AI Agents: models and systems that allow stakeholders to improve their decision making.
Reliable evaluations: curating and benchmarking datasets for trustworthy real-world AI deployment.
Responsible Deployments: working closely with stakeholders to responsibly deploy these systems in the real world.

Want to see the research in motion? See the interactive demos →

KDD 2026Datasets and Benchmarks Track

RELIANCE: Curating and Evaluating Reproductive Health Information on Social Media

Vaibhav Balloli, Laura Peyton Ellis, Vishala Mishra, Alice M Chi, Alex Friedman Peahl, Elizabeth Bondi-Kelly

Devise data curation multi-modal agents to collect and curate reproductive health information from TikTok and work with clinicians to evaluate the accuracy of the content and how LLMs perform in evaluating the content for accuracy and harms.

arXiv Proceedings Project Details →

@inproceedings{balloli2026reliance,
  title          = {{RELIANCE}: Curating and Evaluating Reproductive Health Information on Social Media},
  author         = {Balloli, Vaibhav and Ellis, Laura Peyton and Mishra, Vishala and Chi, Alice M and Friedman Peahl, Alex and Bondi-Kelly, Elizabeth},
  booktitle      = {Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining},
  year           = {2026},
  doi            = {10.1145/3770855.3817447},
  url            = {https://doi.org/10.1145/3770855.3817447},
  eprint         = {2606.18285},
  archivePrefix  = {arXiv},
  primaryClass   = {cs.IR}
}

Interactive Health 2026

PATHFinder Agent for Tailored Prenatal Care

Vaibhav Balloli, Carissa Samuel, Samia Abdelnabi, Alex Peahl, Elizabeth Bondi-Kelly

Interactive LLM agents that converse with pregnant people to provide personalized prenatal care information, planning, and resources to patients, and to assist clinicians.

arXiv Proceedings Project Details →

@inproceedings{balloli2026pathfinder,
  title     = {PATHFinder Agent for Tailored Prenatal Care},
  author    = {Balloli, Vaibhav and Samuel, Carissa and Abdelnabi, Samia and Peahl, Alex and Bondi-Kelly, Elizabeth},
  booktitle = {Proceedings of the 2026 ACM Interactive Health Conference},
  year      = {2026},
  doi       = {10.1145/3786579.3804996},
  url       = {https://doi.org/10.1145/3786579.3804996},
  eprint    = {2607.24768},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CY}
}

FAccT 2026

"Where is this coming from?" Uncovering Trustworthiness Ideals in AI-powered Peripartum Information Seeking

Vaibhav Balloli, Julia Erickson, Xinyi Li, Erin MacMurray van Liemt, Alex Friedman Peahl, Elizabeth Bondi-Kelly

Conduct focus groups to understand the role of AI-enabled information provision in the current health ecosystem and the trustworthiness and governance expectations of stakeholders.

arXiv Proceedings Details →

@inproceedings{balloli2026sociotechnical,
  title          = {``Where is this coming from?'' Uncovering Trustworthiness Ideals in {AI}-powered Peripartum Information Seeking},
  author         = {Balloli, Vaibhav and Erickson, Julia and Li, Xinyi and MacMurray van Liemt, Erin and Friedman Peahl, Alex and Bondi-Kelly, Elizabeth},
  booktitle      = {Proceedings of the 2026 ACM Conference on Fairness, Accountability, and Transparency},
  year           = {2026},
  doi            = {10.1145/3805689.3812277},
  url            = {https://doi.org/10.1145/3805689.3812277},
  eprint         = {2606.10158},
  archivePrefix  = {arXiv},
  primaryClass   = {cs.HC}
}

CV4Animals @ CVPR 2026

SEEK-CBM: Editable and Interpretable Retrieval for Elephant Re-Identification

Antoine Salaün, Vaibhav Balloli, Timm Haucke, Elizabeth Bondi-Kelly, Sara Beery

SEEK-CBM upgrades CHAIR to modern vision architectures and self-supervised techniques to achieve a 497% improvement in performance.

PDF Project Code Talk Details →

@inproceedings{seekcbm2026salaun,
  title     = {{SEEK}-{CBM}: Editable and Interpretable Retrieval for Elephant Re-Identification},
  author    = {Sala{\"u}n, Antoine and Balloli, Vaibhav and Haucke, Timm and Bondi-Kelly, Elizabeth and Beery, Sara},
  booktitle = {6th Workshop on CV4Animals: Computer Vision for Animal Behavior Tracking and Modeling, in conjunction with CVPR 2026},
  year      = {2026}
}

COLING 2025

Oral

Bridging the Gap: Dynamic Learning Strategies for Improving Multilingual Performance in LLMs

Somnath Kumar*, Vaibhav Balloli*, Mercy Ranjit, Tanuja Ganu, Kabir Ahuja, Sunayna Sitaraman, Kalika Bali

A dynamic strategy that selects, per query at run-time, the optimal prompt, LLM, and embedding model to maximize multilingual QA performance, improving non-Latin and low-resource language coverage without any extra training.

Paper Details →

@inproceedings{kumar2025leap,
  title     = {Bridging the Gap: Dynamic Learning Strategies for Improving Multilingual Performance in LLMs},
  author    = {Kumar, Somnath and Balloli, Vaibhav and Ranjit, Mercy and Ganu, Tanuja and Ahuja, Kabir and Sitaraman, Sunayna and Bali, Kalika},
  booktitle = {Proceedings of the International Conference on Computational Linguistics (COLING)},
  year      = {2025},
  note      = {Oral}
}

IJCAI 2024Human-Centered AI track

<5% accept

Are They the Same Picture? Adapting Concept Bottleneck Models for Human-AI Collaboration in Image Retrieval

Vaibhav Balloli, Sara Beery, Elizabeth Bondi-Kelly

CHAIR adapts Concept Bottleneck Models so experts can intervene on intermediate concepts during image retrieval, a setting that matters for wildlife re-identification platforms like ElephantBook. Flexible levels of intervention let domain experts spend effort where it's most useful, achieving genuine human-AI complementarity.

PDF Proceedings Project Code Details →

@inproceedings{balloli2024chair,
  title     = {Are They the Same Picture? Adapting Concept Bottleneck Models for Human-AI Collaboration in Image Retrieval},
  author    = {Balloli, Vaibhav and Beery, Sara and Bondi-Kelly, Elizabeth},
  booktitle = {Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence, {IJCAI-24}},
  editor    = {Kate Larson},
  publisher = {International Joint Conferences on Artificial Intelligence Organization},
  pages     = {7824--7832},
  year      = {2024},
  month     = {8},
  note      = {Human-Centred AI},
  doi       = {10.24963/ijcai.2024/866},
  url       = {https://doi.org/10.24963/ijcai.2024/866}
}

NeurIPS 2023

Chanakya: Learning Runtime Decisions for Adaptive Real-Time Perception

Anurag Ghosh, Vaibhav Balloli, Akshay Nambi, Aditya Singh, Tanuja Ganu

A learned approximate-execution framework for streaming perception. Chanakya jointly considers scene content and system contention to pick run-time decisions (resolution, model, compute) that balance accuracy and latency, beating static and dynamic baselines on both server GPUs and edge devices.

Paper Code Project Details →

@inproceedings{ghosh2023chanakya,
  title     = {Chanakya: Learning Runtime Decisions for Adaptive Real-Time Perception},
  author    = {Ghosh, Anurag and Balloli, Vaibhav and Nambi, Akshay and Singh, Aditya and Ganu, Tanuja},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year      = {2023}
}

Research lineage #

Seven papers, three threads. Each thread starts with the people it is for, grows upward through the work, and ends where that work is headed.

Helping experts steer retrieval

Conservation biologists & rangers
IJCAI 2024 Are They the Same Picture? Adapting Concept Bottleneck...Let experts intervene on concepts mid-search
CV4Animals @ CVPR 2026 SEEK-CBM: Editable and Interpretable Retrieval for...Modern backbones, 497% better retrieval
Research + Deployment To be deployed for the elephant work

Deploying trustworthy health AI systems

Birthing people, clinicians & health workers
FAccT 2026 "Where is this coming from?" Uncovering Trustworthiness...Ask them first: four focus groups, three stakeholder groups
Interactive Health 2026 PATHFinder Agent for Tailored Prenatal CareBuild to that spec: a care-planning agent KDD 2026 RELIANCE: Curating and Evaluating Reproductive Health...Then measure the ecosystem it lives in
Research + Deployment TBD in Michigan

Selecting best system configuration

Read bottom-up: the people each thread serves sit at the root, the papers grow from there, and the top of each thread is where the work is headed.

Tree layout inspired by Philippe Laban.

Interactive research demos #

Animated walkthroughs of how the systems below actually work, drawn from the papers themselves.

SEEK-CBM #

Editable and interpretable retrieval for elephant re-identification.

Read the paper →

PATHFinder #

Interactive agents for tailored prenatal care.

View the project →

Things I keep going back to #

A short, evolving shelf of books and talks I find myself returning to.

📚 Reading

Project Hail Mary
Andy Weir
A problem-solving-as-survival sci-fi novel: first-principles reasoning under extreme constraints.
Deep Work
Cal Newport
On cultivating sustained, distraction-free focus as a competitive advantage for knowledge work.
Principles of Mixed-Initiative User Interfaces
Eric Horvitz (CHI 1999)
The foundational paper on designing systems where humans and automated agents collaborate fluidly.
The Bitter Lesson
Richard Sutton
Short essay arguing that general methods leveraging computation tend to win out over hand-crafted knowledge.

🎬 Talks & videos

The Last Lecture: Achieving Your Childhood Dreams
Randy Pausch
Prof. Randy Pausch's wise, funny lecture on living with intention.
How to Live an Asymmetric Life
Graham Weaver (Stanford GSB Last Lecture)
On doing hard things for decades and betting on outsized, asymmetric outcomes.