I am Panfeng (Gavin) Jiang, currently pursuing an M.S. in Computer Science at Columbia University. I received my B.Eng. in Computer Science from ShanghaiTech University and spent an academic year at the University of California, Berkeley as an exchange student. My interests broadly span machine learning, sequential decision-making, LLMs, and embodied & multimodal AI. I have been fortunate to gain research experience across both theory and applications, working with Dr. Weitong Zhang at the University of North Carolina at Chapel Hill on learning and decision-making under uncertainty, with Dr. Lin Shao at the National University of Singapore on vision-language models and robotic navigation, and with Prof. Laurent Kneip at the Mobile Perception Lab on event-based vision and perception.

📝 Publications

đź“– Education

Aug 2026 – Expected Dec 2027

Columbia University

New York, NY

M.S. in Computer Science

Selected coursework
Databases · LLM-Based Generative AI

Sep 2022 – Jun 2026

ShanghaiTech University

Shanghai, China

B.Eng. in Computer Science

GPA 3.75/4.0 · Rank 11/161

Highlights
National Scholarship · Outstanding Graduate of Shanghai

Aug 2024 – Jun 2025

UC Berkeley

Berkeley, CA

Exchange in Computer Science

GPA 3.8/4.0

Selected coursework
Optimization · Deep Learning · AI · LLM Agents

🏅 Honors and Awards

  • Jun 2026 — Outstanding Graduate of Shanghai.
  • Jun 2026 — Outstanding Graduate of ShanghaiTech University.
  • Jun 2026 — Outstanding Graduate of Dadao College.
  • Dec 2025 — National Scholarship.
  • 2025 — ShanghaiTech Outstanding Student Award (Top 2%).
  • Jun 2025 — Outstanding Winner Award (Top 0.1%), Mathematical Contest in Modeling 2025.
  • Jun 2025 — American Mathematical Society Award, Mathematical Contest in Modeling 2025 (Top 3 of 21,054 teams).
  • Mar 2025 — First Place, Bay Area Data Science Summit at UC Berkeley, endorsed by Wells Fargo.
  • Sep 2024 — Provincial Third Prize, China Undergraduate Mathematical Contest in Modeling.
  • 2023 — ShanghaiTech Merit Student Award (Top 10%).

đź’» Research Experience

Research Assistant · Columbia Creative Machines Lab

Columbia University logo
  • Study whether confidence and correctness can be inferred from internal LLM activations.
  • Train supervisory models on hidden representations to identify unreliable or deceptive responses.
  • Focus on interpretable signals for trustworthy language-model evaluation.

AI Research Intern · RoboScience & NUS Lin Shao Lab

RoboScience logo National University of Singapore logo
  • Led research resulting in a co-first-author paper accepted to ICRA 2026.
  • Built a VLM-based navigation framework that joins high-level reasoning with real-time control.
  • Reached 91.3% navigation success, compared with 28% for the prior state of the art.
View research details

Research Assistant · UNC Chapel Hill, Weitong Zhang Lab

UNC Chapel Hill logo
  • Coauthored a manuscript on variance-aware Langevin Monte Carlo Thompson Sampling.
  • Extended LMC-TS to bandits with action-dependent noise variance.
  • Derived regret bounds connecting finite-step posterior error with exploration efficiency.
View research details

Research Assistant · ShanghaiTech Mobile Perception Lab

Mobile Perception Lab logo
  • Coauthored an ICCV 2025 Highlight paper on asynchronous structure and motion.
  • Developed event-based detection and point-tracking modules for motion estimation.
  • Helped achieve 63 ÎĽs CPU runtime and 10–30% lower velocity error.
View research details

🔍 Projects

May 2025 – Oct 2025 LISN social navigation system

ICRA 2026 · Accepted

LISN: Language-Instructed Social Navigation

  • Introduced a benchmark for instruction-following social navigation.
  • Combined a VLM reasoning loop with real-time costmap and controller modulation.
  • Achieved 91.3% average success across challenging human–robot scenarios.

Project details → · Paper · Video

Nov 2024 – Mar 2025 Asynchronous structure and motion solver

ICCV 2025 · Highlight

A Linear N-Point Solver for Asynchronous Tracks

  • Unified structure and linear-motion estimation across arbitrary timestamps and views.
  • Supported global-shutter, rolling-shutter, event-camera, and mixed sensor tracks.
  • Delivered 63 ÎĽs CPU runtime with 10–30% lower velocity error.

Project details → · Paper · Code

Feb 2025 – May 2025 Sustainable tourism modeling

MCM 2025 · Outstanding Winner & AMS Award

Rebalancing Nature’s Scale

  • Built a coupled ODE model for tourism, environmental pressure, and local revenue.
  • Optimized interventions and tax allocation through simulation and AHP.
  • Placed Top 3 among 21,054 teams in the Mathematical Contest in Modeling.

Project details →

View all projects

đź’Ľ Activities

  • Teaching Assistant, CS101A Data Structures (ShanghaiTech Honors Class), Aug 2025 – Present. Helped launch the university’s first honors data-structures course and developed its syllabus, assignments, and advanced materials.
  • Designer, “1001 Inventions” Exhibition, Jul 2025. Created a mathematically inspired Rubik’s Cube pattern selected by ShanghaiTech’s Humanities Institute.
  • Staff and Referee, Official Rubik’s Cube Competitions, Jul 2023 – Mar 2025. Officiated WCA events in Berkeley, Changsha, and other venues.