Yang Bai (Tony)

Research Scientist at Meta | Machine Learning, Causal Inference, Large-Scale Decision Systems

Resume

Biography

Yang Bai is a Research Scientist at Meta working on large-scale machine learning and decision systems for Instagram Ads. His work spans predictive modeling, causal inference, counterfactual modeling, online experimentation, and scalable ML systems, including models trained on billions of observations and deployed in production.

He received his Ph.D. in Computer Science from the University of Florida, where his research focused on information retrieval, NLP, dense retrieval, and multimodal question answering. His publications include NAACL Findings, COLING, and SIGIR.

SKILLS

Programming: Python, C++, Java, SQL
Machine Learning / Research: Predictive Modeling, Deep Learning, Causal Inference, Counterfactual Modeling, Learning-to-Rank, Retrieval Systems, Online Experimentation
Frameworks: PyTorch, TensorFlow, Transformers, Scikit-learn
Data / Infrastructure: Spark, Linux, Docker, Distributed Training
Domains: Information Retrieval, Ranking Systems, NLP, Multimodal Question Answering

News

[Career] [2025-Present] Working as a Research Scientist at Meta on large-scale machine learning and decision systems for Instagram Ads.
[Paper] [Jan 2025] RAMQA was accepted to NAACL Findings 2025.
[Degree] [Dec 2024] Received a Ph.D. in Computer Science from the University of Florida.
[Paper] [Feb 2024] M3 was published at COLING 2024.
[Paper] [Jul 2023] MythQA was published at SIGIR 2023.

Work Experience

Research Scientist | Meta — Feb 2025-Present
- Built large-scale engagement prediction models for Instagram Feed Ads, trained on billions of user interactions.
- Designed causal inference and counterfactual models to estimate engagement sensitivity under changing ad exposure policies.
- Migrated training pipelines into Meta's NRML production infrastructure and deployed models through large-scale A/B experiments.
- Increased Instagram Feed revenue by approximately 0.6% (~$0.6M/day) while maintaining neutral impact on top-line engagement metrics.
Machine Learning Engineer Intern | Meta — Summer 2024
- Built neural ranking models for Threads Search and generated synthetic relevance labels for training at scale.
- Ran offline evaluations across model architectures to measure search relevance improvements.
Machine Learning Intern | Nokia Bell Labs — Summer 2022
- Developed contrastive-learning dense retrieval for incident analysis spanning 10-100 million log lines per ticket, improving retrieval performance by 16.1% over BM25.
Graduate Researcher | University of Florida — 2019-2024
- Developed multi-hop dense retrieval for open-domain fact verification on FEVER and built resources for multi-answer question answering.
- Built event-centric knowledge graph search for DARPA AIDA; the team ranked #1 in the NIST TAC SM-KBP 2020 evaluation.

Selected Research Projects

RAMQA — A unified retrieval-augmented framework for multimodal question answering. NAACL Findings 2025.
M3 — A multi-task, mixed-objective framework for open-domain multi-hop dense sentence retrieval. COLING 2024.
MythQA — Query-based claim detection through multi-answer open-domain question answering. SIGIR 2023.
DARPA AIDA / GAIA — Event-centric knowledge graph search and hypothesis generation; ranked #1 in NIST TAC SM-KBP 2020.

Selected Publications

RAMQA: A Unified Framework for Retrieval-Augmented Multi-Modal Question Answering
Yang Bai, Christan Grant, and Daisy Zhe Wang. Findings of NAACL, 2025.
M3: A Multi-Task Mixed-Objective Learning Framework for Open-Domain Multi-Hop Dense Sentence Retrieval
Yang Bai, Anthony Colas, and Daisy Zhe Wang. COLING, 2024.
MythQA: Query-Based Large-Scale Check-Worthy Claim Detection through Multi-Answer Open-Domain Question Answering
Yang Bai, Anthony Colas, and Daisy Zhe Wang. SIGIR, 2023.
Can Knowledge Graphs Simplify Text?
Anthony Colas, Haodi Ma, Xuanli He, Yang Bai, and Daisy Zhe Wang. CIKM, 2023.
Improving Rare Tree Species Classification Using Domain Knowledge
Ira Harmon, Sergio Marconi, Ben Weinstein, Yang Bai, Daisy Zhe Wang, Ethan White, and Stephanie Bohlman. IEEE Geoscience and Remote Sensing Letters, 2023.

Research Interests

Machine Learning and Decision Systems
Causal Inference and Experimentation
Ranking and Information Retrieval
Natural Language Processing
Multimodal Question Answering

Education

Ph.D. in Computer Science, 2019-2024
University of Florida, USA
M.S. in Electrical and Computer Engineering, 2016-2018
University of Florida, USA
B.S. in Microelectronics, 2012-2016
Sichuan University, China

Achievements

Gartner Group Graduate Fellowship
CISE Department, University of Florida — 2022 and 2023

#1, NIST TAC SM-KBP 2020
DARPA AIDA / GAIA alternative hypothesis generation evaluation