# Vaibhav Balloli

> Vaibhav Balloli, applied AI researcher and CS Ph.D. candidate at Michigan, building interactive AI systems for health, conservation, and education.

Vaibhav Balloli is an Applied AI Researcher and Computer Science Ph.D. Candidate at University of Michigan, building and deploying safe, reliable interactive AI systems for high-stakes societal domains including healthcare, wildlife conservation, road safety, and education. Research areas: Large Language Models, Reinforcement Learning, Representation Learning, Human-AI Interaction, Agentic Systems, Machine Learning. Previously a Research Fellow and intern at Microsoft Research (HAMS, Vasudha, VeLLM).

This file follows the llms.txt convention (https://llmstxt.org). For structured data (JSON endpoints, markdown alternates, and the full site text in one fetch), see https://profile.vballoli.com/agents/.

## Machine-readable
- [Manifest](https://profile.vballoli.com/agents/index.json): Every machine-readable resource on this site, with media types and field lists. Start here.
- [Full site text](https://profile.vballoli.com/llms-full.txt): Profile, research lineage, systems, all abstracts, topics, news, and CV concatenated into one document.
- [Publications JSON](https://profile.vballoli.com/agents/papers.json): All publications with DOI, arXiv id, authors, venue, abstract, topics, lineage, and BibTeX.
- [Profile JSON](https://profile.vballoli.com/agents/profile.json): Identity, persistent identifiers, education, research experience, and deployed systems.
- [Machine-readable index](https://profile.vballoli.com/agents/): Human-readable documentation of all of the above.

## Publications
- [RELIANCE: Curating and Evaluating Reproductive Health Information on Social Media](https://profile.vballoli.com/papers/reliance/): KDD 2026 (Datasets and Benchmarks Track). Authors: Vaibhav Balloli, Laura Peyton Ellis, Vishala Mishra, Alice M Chi, Alex Friedman Peahl, Elizabeth Bondi-Kelly. DOI: 10.1145/3770855.3817447. arXiv: 2606.18285.
- [PATHFinder Agent for Tailored Prenatal Care](https://profile.vballoli.com/papers/pathfinder/): Interactive Health 2026. Authors: Vaibhav Balloli, Carissa Samuel, Samia Abdelnabi, Alex Peahl, Elizabeth Bondi-Kelly. DOI: 10.1145/3786579.3804996. arXiv: 2607.24768.
- ["Where is this coming from?" Uncovering Trustworthiness Ideals in AI-powered Peripartum Information Seeking](https://profile.vballoli.com/papers/sociotechnical-design/): FAccT 2026. Authors: Vaibhav Balloli, Julia Erickson, Xinyi Li, Erin MacMurray van Liemt, Alex Friedman Peahl, Elizabeth Bondi-Kelly. DOI: 10.1145/3805689.3812277. arXiv: 2606.10158.
- [SEEK-CBM: Editable and Interpretable Retrieval for Elephant Re-Identification](https://profile.vballoli.com/papers/seek-cbm/): CV4Animals @ CVPR 2026. Authors: Antoine Salaün, Vaibhav Balloli, Timm Haucke, Elizabeth Bondi-Kelly, Sara Beery.
- [Bridging the Gap: Dynamic Learning Strategies for Improving Multilingual Performance in LLMs](https://profile.vballoli.com/papers/leap/): COLING 2025 [Oral]. Authors: Somnath Kumar*, Vaibhav Balloli*, Mercy Ranjit, Tanuja Ganu, Kabir Ahuja, Sunayna Sitaraman, Kalika Bali. arXiv: 2405.18359.
- [Are They the Same Picture? Adapting Concept Bottleneck Models for Human-AI Collaboration in Image Retrieval](https://profile.vballoli.com/papers/chair/): IJCAI 2024 (Human-Centered AI track) [<5% accept]. Authors: Vaibhav Balloli, Sara Beery, Elizabeth Bondi-Kelly. DOI: 10.24963/ijcai.2024/866.
- [Chanakya: Learning Runtime Decisions for Adaptive Real-Time Perception](https://profile.vballoli.com/papers/chanakya/): NeurIPS 2023. Authors: Anurag Ghosh, Vaibhav Balloli, Akshay Nambi, Aditya Singh, Tanuja Ganu. DOI: 10.52202/075280-2429.

## Systems and deployments
- **HAMS: automated driver licence testing** (Microsoft Research India, 2021 – 2022, status: Deployed): 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.
- **PATHFinder: prenatal care planning agent** (University of Michigan, 2025 – 2026, status: In progress): 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.
- **CHAIR & SEEK-CBM: steerable wildlife re-identification** (University of Michigan · MIT, 2023 – 2026, status: In progress): 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.
- **VeLLM: multilingual LLMs for the global majority** (Microsoft Research India, 2022 – 2023, status: Prototyping): 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.

## Research lineage
- *Helping experts steer retrieval* for conservation biologists & rangers. Are They the Same Picture? Adapting Concept Bottleneck Models for Human-AI Collaboration in Image Retrieval → SEEK-CBM: Editable and Interpretable Retrieval for Elephant Re-Identification. Stage: Research + Deployment. Deployment: To be deployed for the elephant work.
- *Deploying trustworthy health AI systems* for birthing people, clinicians & health workers. "Where is this coming from?" Uncovering Trustworthiness Ideals in AI-powered Peripartum Information Seeking → PATHFinder Agent for Tailored Prenatal Care and RELIANCE: Curating and Evaluating Reproductive Health Information on Social Media. Stage: Research + Deployment. Deployment: TBD in Michigan.
- *Selecting best system configuration* for adaptive systems. Chanakya: Learning Runtime Decisions for Adaptive Real-Time Perception → Bridging the Gap: Dynamic Learning Strategies for Improving Multilingual Performance in LLMs. Stage: Research.

## Topics
- [AI for Social Impact](https://profile.vballoli.com/topics/social-impact/): Deployments where the measure of success is whether something changed for the people involved: road safety testing across India, elephant re-identification for conservation partners, prenatal care planning with clinicians in the loop. The recurring lesson is that the deployment constraints reshape the model, not the other way round.
- [AI for Health](https://profile.vballoli.com/topics/healthcare/): Research on AI in peripartum and reproductive health, spanning the full arc from formative study to deployed system to benchmark: what birthing people, clinicians, doulas, and community health workers actually need from an information tool, an agent built to those requirements, and a dataset for checking whether the surrounding information ecosystem is accurate.
- [Large Language Models](https://profile.vballoli.com/topics/llms/): Work on making large language models useful when the stakes are real: choosing prompts, models, and embeddings at run time rather than fixing them in advance, building conversational agents that gather context before they advise, and measuring where frontier models still fall short. The thread running through it is that an LLM is a component in a system, not the system itself.
- [Computer Vision](https://profile.vballoli.com/topics/vision/): Vision systems designed to be steered by the people who use them. This includes concept-bottleneck retrieval for wildlife re-identification, where a domain expert can intervene on intermediate concepts mid-search, and adaptive streaming perception that trades accuracy against latency at run time based on scene content and system contention.
- [Evaluation & Benchmarks](https://profile.vballoli.com/topics/evals/): Datasets and evaluation protocols built with the domain experts who have to trust the result. Rather than scoring models on proxies, this work curates expert-annotated ground truth (clinicians reviewing health claims, rubrics written by practising obstetricians) and asks whether language models can stand in for that judgement. Usually the answer is partly, and the interesting part is where the gap sits.

## Links
- [Homepage](https://profile.vballoli.com/): About, news, systems, publications, research lineage, and reading list
- [Curriculum Vitae](https://profile.vballoli.com/cv/): Education, research experience, publications, awards, and service ([PDF](https://profile.vballoli.com/uploads/Vaibhav_Balloli_CV.pdf))
- [GitHub](https://github.com/vballoli): Code and open-source projects
- [Google Scholar](https://scholar.google.com/citations?user=-6i7e7sAAAAJ&hl=en): Publication list and citations
- [ORCID](https://orcid.org/0000-0003-1036-9672): Persistent researcher identifier
- [Semantic Scholar](https://www.semanticscholar.org/author/1564592237): Publication record
- [DBLP](https://dblp.org/pid/348/6962): Computer science bibliography
- [LinkedIn](https://www.linkedin.com/in/vaibhavballoli): Professional profile
- [Twitter / X](https://twitter.com/v_balloli): Updates
- [Email](mailto:balloli.vb@gmail.com): Contact

## Optional
- [Friends](https://profile.vballoli.com/friends/): Friends of the author and their websites
- [Reading list](https://profile.vballoli.com/#reading): Books and talks the author returns to
- [News](https://profile.vballoli.com/#news): Dated announcements, also included in the full site text

## About
- Name: Vaibhav Balloli
- Role: Applied AI Researcher and Computer Science Ph.D. Candidate, University of Michigan
- Advisor: Elizabeth Bondi-Kelly
- Research: Large Language Models, Reinforcement Learning, Representation Learning, Human-AI Interaction, Agentic Systems, Machine Learning
- Previously: Microsoft Research (Research Fellow; Research Intern)
- ORCID: 0000-0003-1036-9672