I'm harnessing AI to transform document insights and data extraction at Pfizer, driving impactful change in healthcare.
Explore my projects that utilize AI for data insights and document analysis, showcasing my skills and impact in healthcare.
.jpg)
Pfizer · Oct 2026 · ✅ Verified by Extern
Prototype AI-powered document intelligence with Pfizer—using OCR, LLMs, and RAG to automate real enterprise PDF workflows and build a standout portfolio project.
Analyzed investment flows across the hydrogen value chain (upstream, midstream, downstream) to identify capital concentration patterns and evaluate alignment with emission reduction potential. Cleaned Pitchbook data and applied Logit/Probit models to identify key drivers of capital allocation across gray, blue, and green hydrogen technologies. Examined financing challenges facing Chinese clean-tech startups, analyzed public sector risk guarantee mechanisms, and proposed institutional recommendations to expand financing channels.
I'm harnessing AI to transform document insights and data extraction at Pfizer, driving impactful change in healthcare.
I am currently an extern at Pfizer, focusing on AI-driven document insights and data extraction. My work combines my passion for technology with practical applications in the healthcare sector.
Externships
Pfizer Advanced: AI-Powered Document Insights & Data Extraction Externship
Completed · Oct 2026
Experience
Intern, Experiment and Analytical Testing Center
Jiangsu Provincial Academy of Environmental Science · Jun-Aug 2024
Education
Duke University
Master of Environmental Policy · Class of 2027
University of Toronto
Honours Bachelor of Science with Distinction in Environmental Science and Statistics · Class of 2025
Skills
✅ Verified by Extern
Prototype AI-powered document intelligence with Pfizer—using OCR, LLMs, and RAG to automate real enterprise PDF workflows and build a standout portfolio project.
The work processed vendor and pharmaceutical SDF PDFs through CV/OCR and NLP pipelines to classify document types, extract batch/retest numbers and date fields, and log OCR failures and layout issues. Deliverables included extracted batch and retest numbers, Effective/Manufacturing/Expiration dates, file timestamps, an OCR error log, a page-level JSON map linking pages to doc_id and doc_type, OCR
I extracted structured metadata (batch and retest numbers, Effective/Manufacturing/Expiration dates), recorded file creation and modification timestamps, logged OCR failures and layout issues, compared Tesseract, PaddleOCR, and EasyOCR, produced a page-level JSON map, and deployed a Gradio RAG chatbot with multiple retriever configurations.
Analyzed investment flows across the hydrogen value chain (upstream, midstream, downstream) to identify capital concentration patterns and evaluate alignment with emission reduction potential. Cleaned Pitchbook data and applied Logit/Probit models to identify key drivers of capital allocation across gray, blue, and green hydrogen technologies. Examined financing challenges facing Chinese clean-tech startups, analyzed public sector risk guarantee mechanisms, and proposed institutional recommendations to expand financing channels.
Investigated the "Wu Wei Er Zhi" (effortless action) approach, a method of tea cultivation that uses no fertilizers or pesticides and relies on biodiversity for pest control, while tracing the historical factors that preserved ancient tea trees. Conducted surveys and in-depth interviews with 20+ farmers, intermediaries, and village officials, collecting primary data on industry status and income structures. Quantified the premium value generated by the natural farming model and identified structural bottlenecks such as “middleman dependency” that destabilize farmer incomes. Extracted ESG value from environmental, social, and governance dimensions, providing a replicable framework for sustainable ecological agriculture.
Identified a critical policy gap: the current regulatory framework covers less than 90% of SMEs, with 30% of firms achieving below 70% exhaust gas collection efficiency, directly undermining the regional ozone pollution control target. Surveyed 3 major industrial clusters, evaluated 3 policy alternatives, and conducted quantitative comparisons of outcomes. Authored a 21-page policy white paper, proposing two cross-departmental coordination strategies and actionable recommendations for macro-level environmental governance.
Independently designed and executed leaching experiments on cup lids from seven major brands; analyzed chemical additives using LC-MS and implemented Exposure-Activity Ratio (EAR) models with R for data visualization. Conducted a review of 20+ studies, identifying gaps in current regulations regarding secondary microplastic control. Analyzed microplastic applications in the PPCP industry, uncovered greenwashing in cosmetic product formulations, and proposed a multi-stakeholder governance framework involving consumers, corporations, and government.
Processed FAO global water resource data using R and applied statistical methods to identify high-consumption industries and the top 10 water-stressed countries. Developed interactive dashboards in Tableau (maps, bar charts, bubble charts) to visualize water allocation across agriculture, industry, and domestic sectors, along with spatial patterns of desalination capacity. Delivered data-driven recommendations to optimize resource efficiency and inform strategies for water-scarce regions.
与全国防鸟撞行动网络合作，独立设计并执行鸟类尸体移除实验，量化清道夫活动对鸟撞监测的干扰。运用对比分析与实验数据，发现 91%模拟尸体被清道夫（主要为校园流浪猫）移除，71%清除发生在夜间，证实传统日间巡查方法存在系统性低估。利用 Tableau 进行空间与物种分析，识别高风险建筑与易感物种，为精准干预提供数据支撑。提出引入“清道夫校正因子”估算真实死亡率，并向校园管理部门提交“鸟类友好型建筑设计”建议。
