Yicheng Tao 陶毅诚

Pittsburgh, PA · eric.taoyicheng@gmail.com(Personal) · yt2@andrew.cmu.edu(CMU)

I'm a graduate student in the Master of Science in Intelligent Information Systems (MIIS) program at Carnegie Mellon University's School of Computer Science (SCS), within the Language Technologies Institute (LTI), guided by Prof. Lei Li.

Previously, I completed my Bachelor's degree in Computer Science and Technology at Southern University of Science and Technology (SUSTech) in Shenzhen, China. I conducted research under the supervision of Prof. Xuan Song and Prof. Yepang Liu.
My research interests include LLM reasoning, code generation, agentic search, and multimodal spatiotemporal data mining.

Last updated: September 2026.

Research

Retrieval-Augmented Code Generation: A Survey with Focus on Repository-Level Approaches (Link)

Lead Contributor · TMLR · 40+ Citations

A comprehensive survey of repository-level code generation, covering generation and retrieval methods, training paradigms, agent design, and evaluation protocols.

Jun 2025 - May 2026

Enhancing LLMs for Mobility Analytics with Semantic Location Tokenization (Link)

Lead Contributor · KDD 2025 · Nanyang Technological University

Introduced discrete location tokens and a multi-task fine-tuning strategy that helps LLMs learn mobility semantics, outperforming prior methods across multiple datasets by 38.5%, 40.6%, and 14.7%.

Sep 2024 - Apr 2025

PlanGPT (Link)

ACL 2025 Industry · Peking University

Built data and retrieval components for an urban-planning language model, including 3,000+ CEVAL questions, LoRA fine-tuning, and Plan-Emb, a specialized embedding model that improved over the previous state of the art by 2%.

Feb 2024 - Mar 2024

FANNO (Link)

ACL 2025 Findings

Developed an automated instruction-annotation framework with pre-screening, seed generation, augmentation, and response generation, outperforming the previous best method by 41.7%.

Mar 2024 - Sep 2024

LLM-enhanced Causality-aware Traffic Accidents Prediction (Link)

IEEE T-ITS 2025 · University of Tokyo

Designed a multi-source dataset-alignment framework for causal modeling and demonstrated an average 1.51% improvement over existing methods.

Aug 2024 - Jul 2025

Working Experience

Bosch

LLM & Agentic AI R&D Intern · Mentors: Lukas Lange, Jun Araki
  • Engineered retrieval and evaluation pipelines for coding agents, enabling context extraction, reuse, and iterative improvement.
  • Built a learning-to-rank code retrieval system with fine-grained feedback that achieved the highest MAP in internal evaluations.
  • Reduced dollar cost by 59.9% with context compression and ranking, then cut token usage by a further 9.8% with trajectory-based reranker training; the work was submitted to NAACL.
Sunnyvale, CA May 2026 - Aug 2026

Gaorong Capital (Link)

LLM Research Intern · Mentors: Bin Zhang, Panlin Chen
  • Rebuilt the retrieval pipeline with hybrid RAG and local reasoning models, improving recall by 27% over the baseline.
  • Developed the firm's first internal LLM research assistant using MCP tools, multi-agent search, Deep Research, iterative QA, and the STORM framework.
Shenzhen, China Mar 2025 - Jun 2025

LocationMind Inc. (Link)

Research Intern · Mentor: Zipei Fan
  • Designed a lightweight geolocation agent combining GeoCLIP retrieval with VLM reasoning for grounded candidate verification.
  • Applied GRPO to learn when to trust or override retrieved hypotheses, doubling Street@1km accuracy from 6.4% to 13.8%.
Tokyo, Japan & Remote Jul 2024 - Apr 2026

Behavioral and Spatial AI Lab, Peking University

Research Assistant

Assisting in the development of the nation's first urban planning large language model and conducting experiments.

Shenzhen, China Feb 2024 - Mar 2024

SuGaoxin Digital Technology Co., Ltd. (Link)

Java Backend Intern

Developed middleware for efficient management of Modbus, BACnet, and Honeywell edge devices.

Suzhou, China Aug 2023

Education

Carnegie Mellon University

Language Technologies Institute, School of Computer Science
Master of Science in Intelligent Information Systems (MIIS)
QPA: 4.00/4.00 · Advisor: Lei Li
Relevant coursework: Search Engine, Multimodal Machine Learning, Advanced Natural Language Processing, LLM Application
August 2025 - December 2026

Southern University of Science and Technology

Dept. of Computer Science and Engineering
Bachelor of Engineering in Computer Science and Technology
GPA: 3.79/4.00 · Top 18% (Junior year: 8/195)
Graduated with Outstanding Graduate and Thesis awards
Advisors: Xuan Song (Research), Yepang Liu (Thesis)
August 2021 - July 2025

Skills

Programming Languages & Tools
    • C++ Icon
LLM-related
  • Fine-tuning (SFT, LoRA)
  • Agentic Workflows
  • Retrieval-Augmented Generation (RAG)
  • Prompt Engineering
  • Reinforcement Learning
Data Science-related
  • PyTorch, TensorFlow, scikit-learn, pandas, NumPy, Matplotlib
Full Stack
  • Experienced in both front-end and back-end development.
  • Successfully completed full-score projects such as "Chess Game," "Course Project Management System," "SUSTech Event Center," and "StackOverflow Visualization," all selected to be showcased in class.
Languages
  • Proficient in Mandarin, fluent in English, and basic knowledge of Japanese.

Awards & Certifications

  • 1 st Class - SUSTech Merit Student Scholarship - 2024
  • SUSTech Merit Student Scholarship - 2023, 2022
  • SUSTech Outstanding Student Award - 2024, 2023
  • 3 rd Place - Guangdong Contemporary Undergraduate Mathematical Contest in Modeling - 2023
  • MIT Machine Learning Winter Course - February 2023
    Base SPOC: Machine Learning, Modeling, and Simulation Principles, focusing on mathematics and knowledge in ML.
  • NUS Deep Learning Summer Workshop - July 2023
    SWS3009 Robotics & Deep Learning, completed tasks including cat classification (using YOLOv5, MobileNet), and video action recognition with MediaPipe and LSTM model.
  • NTU Visiting Research - September 2024 to January 2025
    Conduct research on spatial-temporal urban analysis combined with LLM. Supervised by Cong Gao, Yile Chen.

Services

Community: Actively engaged in volunteer work, accumulating 64 hours of service to date.

Campus: Served as the Director of the Design Division of the College Propaganda Center from 2022 to 2023, and as a member of the WeChat Department in the school's student news agency.

Activity: Participated as a committee member in the ”Innovation and Entrepreneurship Day Training Camp” organized by the Employment Guidance Center at SUSTech and was recognized as an outstanding participant.