Tianle Zhou

Research assistant in systems for AI agents and data systems.

Tianle ZHOU

Tel: +1 347-538-8158 Email: zhout060@gmail.com LinkedIn: tianle-zhou-99a245221 GitHub: dalel213

Education Background

Columbia University — Bachelor of Computer Science (GPA: 3.95/4.3)
Sep 2023 – May 2026 · New York, NY
Relevant Courses: Advanced Programming (C++), Operating Systems (UNIX), Machine Learning, Analysis of Algorithms, Cloud Computing, Advanced Software Engineering, User Interface Design, Database Systems
Scholarship: Dr Herman Hu Scholarships for Joint Degree Programmes (350k HKD, 3–7 awardees)

City University of Hong Kong — Computational Finance & Financial Technology (GPA: 3.87/4.3), Top 5%
Sep 2021 – May 2023 · Hong Kong, CN
Relevant Courses: Optimization Methods, Security Analysis & Portfolio Management, Database Systems, Data Structures, Computer Programming, Differential Equations, Discrete Mathematics, Multivariable Calculus
Scholarship: HKSAR Government Scholarship Fund — Talent Development Scholarship

Research Experiences

Columbia University — Task decomposition for agents
Research Assistant, DAP Lab; Supervisors: Prof. Eugene Wu
Sep 2024 – Current · New York

  • Developed a constraint graph modeling approach to capture complex state dependencies (e.g., structured queries), enabling principled reasoning for LLM-based agents.
  • Applied reduction theory and constraint process techniques to formalize boundaries between solvable vs. intractable dependency problems.
  • Deployed the algorithm to text-to-SQL tasks (SPIDER benchmark), achieving significant improvements (~40% enhancement) over baseline LLM agents.

Columbia University — System for agents
Research Assistant, DAP Lab; Supervisors: Prof. Eugene Wu, Prof. Kostis Kaffes
Sep 2024 – Current · New York

  • Designed a system-native exploration framework for LLM agents using Monte Carlo Tree Search (MCTS), adapting reinforcement learning techniques to system-level automation tasks.
  • Benchmarked and optimized snapshotting and restoration pipelines (container checkpointing, state restoration), reducing exploration rollout time by an average of 60% under constrained resources.
  • Conducted experiments on coding and file-management tasks, showing exploration improves success rates vs. single-shot prompting.

City University of Hong Kong — Research Assistant
Global ESG Analytical Lab; Supervisor: Prof. Rengong Zhang
Dec 2023 – Feb 2024

  • Engineered a research-level ESG data disclosure platform with Azure deployment.
  • Employed Vue and FastAPI across front-end and back-end to boost iteration speed and responsiveness.
  • Optimized data visualization with ECharts for dynamic interactions.
  • Utilized ontology mapping to boost ESG research with NLP (Sentence Transformer with BERT).
  • Built an end-to-end pipeline for data augmentation integrating HNSW for efficient ANN search.

Internship Experiences

Sara Technologies — Software Engineer Intern
Sep 2024 – May 2025 · New York

  • Built a pediatric speech improvement tool using NLP to correct children’s pronunciation.
  • Ensured cross-platform compatibility across tablets, iPhones, and OS versions.
  • Used React Native and Figma for front-end; Amplify and AWS for cloud services.
  • Technologies: React Native, Figma, AWS Amplify, AWS, cross-platform development, NLP.

NomadFi — Software Engineer Intern
Jul 2024 – Aug 2024 · New York

  • Configured local training with Docker and deployed via AWS Lambda with Amazon ECR.
  • Designed and maintained a back-end relational database using MySQL.
  • Employed PPS analysis and SMOTEENN for feature engineering and robust modeling.
  • Enabled API connectivity between front-end and back-end on AWS EC2.
  • Technologies: Docker, AWS Lambda, Amazon ECR, MySQL, PPS, SMOTEENN, AWS EC2, REST APIs.

Publications

See the dedicated Publications page. Each entry includes a reserved teaser image.

Project Experiences

National Champion — CITI Financial Application Competition
Dec 2022 – Jun 2023 · SQL, PyTorch

  • Built a cost forecasting model integrating ESG and sentiment signals aligned to sector benchmarks.
  • Automated data collection and analysis from FTSE Russell and Refinitiv.
  • Supported NLP tasks (BiLSTM, FinBERT) and back-end data via Bloomberg/Refinitiv APIs.
  • Outcome: National Outstanding Award (first overall) and ESG Special Award.

Technical Skills

  • Languages: SQL, C/C++, Python, Java, JavaScript, HTML/CSS, Spring Boot, Flask API, TypeScript
  • Systems: Linux commands, Linux kernel, TCP/UDP, iOS Development, Android Development