Heshan Chen 陈河杉

Institute of Carbon Neutrality & School of Economics, Peking University

PI: Professor Ji Xi, Department of Resource and Environmental Economics

• Constructed a global biodiversity legislation corpus (~150k documents, 190+ jurisdictions) via automated scraping, deduplication, and multilingual retrieval.
• Designed a human-AI collaborative classification framework on a fine-tuned GPT-OSS model, implementing a multi-label policy taxonomy pipeline for CBD, IPBES, and Kunming-Montreal GBF targets.
• Built an end-to-end quality assurance layer with calibrated confidence scoring, entropy-based uncertainty flagging, and structured exclusion reasoning.

Graduate Researcher at Lamont-Doherty Earth Observatory, Columbia Climate School

PI: Professor Beizhan Yan & National Oceanic and Atmospheric Administration (NOAA)

• Developing AI algorithms for autonomous plastic detection using a solar-powered catamaran with visible light and hyperspectral cameras; designing multi-sensor data fusion pipelines for real-time classification.
• Conducting spectral characterization of plastic polymers under hyperspectral imaging; designed a custom NIR LED illumination system to enhance signal-to-noise ratio in field deployments.

Graduate Researcher at Center for Sustainable Urban Development, Columbia Climate School

PI: Professor Jacqueline Klopp, Professor Ebru Gencer, Professor Andrew Kruczkiewicz

• Lead development of a multi-hazard digital-twin platform for post-earthquake recovery, combining satellite imagery, structural inventories, and community feedback; built AI-driven change detection for tracking building, road, and vegetation recovery.
• Design geospatial analytics workflows fusing GIS, remote-sensing time series, climate projections, and AI agent architectures; developed CNN-based landslide detection using SAR imagery.
• Co-develop flash-flood modeling components; integrated compound hazard layers including flood, heat stress, wildfire, air quality, and drought.

Graduate Researcher at Lamont-Doherty Earth Observatory, Columbia Climate School

• Analyzed cumulative impacts of 13 mainstream dams on water level, discharge, and sediment using 65 years of observational data, revealing weakening seasonal flow patterns.
• Developed a baseline-pulse framework to quantify upstream-downstream distance effects; constructed flow prediction models for independent hydrological early warning.

Boston University Department of Electrical and Computer Engineering Featured Project

PI: Professor Osama Alshaykh, Professor Alan Pisano, Professor Ryan Lagoy

• Led design of a multimodal AI system for urban plant health using edge robotics, vision-language models, and cloud infrastructure.
• Implemented YOLOv12x for segmentation, EdenScopeVL for zero-shot disease classification, and EdenScopeVL-CoT reasoning agent integrating GPS, weather, and soil data.
• Built cross-platform applications (macOS, web, mobile) with GIS visualization, live inference, and diagnostic reporting.

AI Mathematical Olympiad

• Distilled Qwen2.5 with DeepSeek R1 using AWQ, fine-tuned with GRPO reinforcement learning for improved mathematical reasoning.
• Implemented chain-of-thought reasoning, self-consistency voting, and optimized sampling strategies to maximize accuracy.

New York State Energy Research and Development Authority (NYSERDA)

“Computing the Costs & Benefits: Data Centers, Grid Policy, and Clean Energy in New York”

• Analyze data center projects in the NYISO interconnection queue and assess their projected impacts on New York’s grid capacity, electricity pricing, and progress toward state clean energy targets.
• Lead research on data center waste heat recovery feasibility, evaluating on-site thermal reuse, district heating integration, and thermal energy storage options to inform state-level energy efficiency policy.
• Lead development of economic policy recommendations addressing grid optimization strategies, rate impact analysis, and clean energy investment trade-offs for new large-scale computing infrastructure in New York State.

Columbia University School of International and Public Affairs

• Co-led a ten-member policy research team to assess zoonotic spillover risks from wildlife trade and live animal markets across New York State; conducted legislative analysis and coordinated interagency framework design integrating the Department of Environmental Conservation, Department of Health, and Department of Agriculture and Markets.
• Developed data-driven policy recommendations including prohibited-species criteria, real-time surveillance networks, and One Health-aligned early warning systems to reduce zoonotic risk statewide.

Engineering Intern, Intelligent Manufacturing R&D Center

• Developed lithium-ion battery SOH prediction algorithms leveraging voltage-current time-series analysis to model capacity degradation, directly informing BMS optimization and safety protocol upgrades.

Electrical and Control Engineer

• Programmed PLC-based servo control systems for Tesla 4680 cell winding lines and built a custom C-based MES platform integrating deep learning models with CCD vision data, improving yield detection accuracy by 10%.

Information Technology & Pipeline Intern

Software Engineering Intern, Ride-Hailing Software Division