Tianyi Peng

Tianyi Peng Assistant Professor at Columbia University Decision, Risk, and Operations Division Data Science Institute Email: tianyi.peng [at] columbia [dot] edu [CV] [GitHub] [linkedin]

About Me

I am broadly interested in AI for decision-making and envision a future where AI agents coexist with humans to enable complex task automation and decision support. My research themes include:

• Ensuring the safe and reliable deployment of AI agents in real-world applications, with continuous alignment and adaptation to business and societal objectives.

• Reducing the operational costs of AI systems to enable scalable, low-latency deployment across diverse environments.

• Leveraging AI to simulate human and system behavior (known as “digital twin”), enabling cost-effective, scalable experimentation and testing.

• Optimizing large-scale decision-making systems through AI and reinforcement learning.

I am a founding member of Cimulate.AI, developing Customer-GPT for e-commerce. We are hiring!

I've collaborated with organizations like Anheuser-Busch InBev, TikTok, and Liberty Mutual. My work with Anheuser-Busch InBev on developing an innovative experimentation platform earned the Daniel H. Wagner Prize in 2022 (INFORMS news). My research has also received the APS Best Student Paper Prize and RMP Jeff McGill Student Paper Award.

I hold a Ph.D. from MIT and a Bachelor's in Computer Science from Tsinghua University (Yao Class).

Publications

• Throughput-Optimal Scheduling Algorithms for LLM Inference and AI Agents
  Yueying Li*, Jim Dai*, Tianyi Peng
  

• How Well do LLMs Compress their Own Chain-of-Thought? A Token Complexity Approach
  Ayeong Lee*, Ethan Che*, Tianyi Peng
  

• Data Mixture Optimization: A Multi-fidelity Multi-scale Bayesian Framework
  Tzu-Ching Yen, Andrew Wei Tung Siah, Haozhe Chen, C. Daniel Guetta, Tianyi Peng, Hongseok Namkoong
  

• LLM Generated Persona is a Promise with a Catch
  Ang Li, Haozhe Chen, Hongseok Namkoong, Tianyi Peng
  

• Differences-in-Neighbors for Network Interference in Experiments
  With Naimeng Ye and Andrew Zheng
  

• QGym: Scalable Simulation and Benchmarking of Queuing Network Controllers
  Haozhe Chen, Ang Li, Ethan Che,Tianyi Peng, Jing Dong, Hongseok Namkoong
  

  • NeurIPS 2024, Dataset and Benchmarks Track

  • [Code]

• Speeding up Policy Simulation in Supply Chain RL
  With Vivek F. Farias, Joren Gijsbrechts, Aryan Khojandi, Andrew Zheng
  

  • Submitted

• Correcting for Interference in Experiments: A Case Study at Douyin
  With Vivek F. Farias, Andrew Zheng, Hao Li, Huawei Zhang, and Xinyuyang Ren
  

  • RecSys 2023

• Markovian Interference in Experiments
  With Vivek F. Farias, Andrew A. Li, and Andrew Zheng
  Under review in Management Science
  

  • Winner, Applied Probability Socity (APS) Best Student Paper Prize 2022

  • Winner, Jeff McGill Student Paper Award 2022

  • NeurIPS 2022, Oral (~2%), [Video in Chinese]

• Generalized Synthetic Control for TestOps at ABI: Models, Algorithms, and Infrastructure
  With Luis Costa, Vivek F. Farias, Patricio Foncea, Jingyuan (Donna) Gan, Ayush Garg, Ivo Rosa Montenegro, Kumarjit Pathak, and Dusan Popovic
  INFORMS Journal on Applied Analytics
  

  • Winner, Daniel H. Wagner Prize 2022

  • [pdf]

• Learning Treatment Effects in Panels with General Intervention Patterns
  With Vivek F. Farias and Andrew A. Li
  

  • Finalist, MSOM Best Student Paper Prize 2022

  • NeurIPS 2021, Oral (~0.6%)

  • [A Python Package for Causal Inference in Panels]

• Fixing Inventory Inaccuracies At Scale
  With Vivek F. Farias and Andrew A. Li
  Accepted by Manufacturing & Service Operations Management
  

  • MSOM Supply Chain Management SIG 2022

  • ICML 2021
    

• Synthetically Controlled Bandits
  With Vivek F. Farias, Ciamac Moallemi, and Andrew Zheng
  Major revision in Management Science
  

  • MSOM Service SIG 2022

• Uncertainty Quantification For Low-Rank Matrix Completion With Heterogeneous and Sub-Exponential Noise
  With Vivek F. Farias and Andrew A. Li
  

  • AISTATS 2022
    

• The Limits to Learning a Diffusion Model
  With Jackie Baek, Vivek F. Farias, Andreea Georgescu, Retsef Levi, Deeksha Sinha, Joshua Wilde, and Andrew Zheng
  Accepted by Management Science

  • Finalist, Post-Pandemic Supply Chain and Healthcare Management Best Paper

  • EC 2021
    

Publications in Quantum (A previous life..)

• Quantum Queuing Delay
  Wenhan Dai, Tianyi Peng, Moe Win
  IEEE Journal on Selected Areas in Communications (IEEE-JSAC), vol. 38, no. 3, pp. 605-618, 2020
  [link], [pdf]

  • Preliminary version: Queuing Delay for Quantum Networks
    International Conference on Computing, Networking and Communications (ICNC 2020)

• Optimal Remote Entanglement Distribution
  Wenhan Dai, Tianyi Peng, Moe Win
  IEEE Journal on Selected Areas in Communications (IEEE-JSAC), vol. 38, no. 3, pp. 540-556, 2020
  [link], [pdf]

  • Preliminary version: Optimal Protocols for Entanglement Swapping and Distribution
    Best Paper Award, International Conference on Computing, Networking and Communications (ICNC 2020)
    

• Simulating Large Quantum Circuits on a Small Quantum Computer
  Tianyi Peng, Maris Ozols, Aram Harrow, Xiaodi Wu
  Physical Review Letters 125, 150504 (2020) (PRL)
  [link] [arxiv]

• Efficient and Robust Physical Layer Key Generation
  Tianyi Peng, Wenhan Dai, Moe Win
  Military Communications Conference 2019 (MILCOM 2019)
  [link]

• Remote State Preparation for Multiple Parties
  Wenhan Dai, Tianyi Peng, Moe Win
  IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2019), Invited Paper
  [pdf], [link]

• Quantum Uncertainty Relation of Coherence
  Xiao Yuan, Ge Bai, Tianyi Peng, Xiongfeng Ma
  Physical Review A 96 (3), 032313 (2017)
  [link] [arxiv]

• Tight Detection Efficiency Bounds of Bell Tests In No-signaling Theories
  Zhu Cao, Tianyi Peng (co-first author)
  Physical Review A 94, 042126 (2016)
  [link] [arxiv]