Pengcheng Lu

About Pengcheng Lu

Most analysts can tell you what happened. I focus on why — and whether the insight will actually hold up when someone makes a real decision based on it. I\'m a UW Information Management graduate student with 2+ years of industry experience across telecom, manufacturing, and finance. What sets me apart is that I approach business problems the way a researcher does: I\'m not satisfied with correlations — I want to isolate actual causation and translate that into something actionable. What this looks like in practice: At China Unicom, I segmented 1M+ user records, identified video streaming as the key retention driver, and restructured a pricing strategy that captured new subscribers — beating acquisition targets by 25%. At China Tower, I traced a 15% billing discrepancy across base stations to a data governance gap, then built protocols that accelerated settlement cycles by 5 days. In my current independent research, I apply the same mindset to a larger-scale problem: using matched-control and event-study causal models on 20.8M Seattle Public Library checkout records to quantify how commercial ranking signals reshape public consumption behavior. I\'ve also designed and run a full RCT on AI disclosure labels — finding that cosmetic transparency measures systematically shift cognitive burden without resolving underlying asymmetries. The common thread: I build things that hold up under scrutiny, whether that\'s a pricing recommendation, a data pipeline, or a causal model. Technical toolkit: Python (Polars, Pandas, RapidFuzz) · SQL · Stata · Tableau · Power BI · A/B Testing · Causal Inference · Experimental Design · Entity Resolution Seeking Summer 2026 internships — Data Analyst, Product Analyst, or Product Manager roles.

Experience

1. Research Analyst

   University Of Washington Information School

   Feb 2026 — Present · Seattle, WA, US

   Identified a +35% structural borrowing advantage for NYT Bestseller-ranked titles over matched controls, quantifying the measurable reach of commercial ranking systems into public cultural consumption• Engineered an asynchronous entity resolution pipeline (Python, Polars, RapidFuzz) to match and clean 20.8M records across disparate library catalog systems — reducing false match rate while processing 1GB+ data• Designed matched-control and event-study causal models to isolate ranking signal effects from confounding factors such as genre, publication date, and baseline popularityFindings submitted to ASIS & T 2026; contributing to policy discussions on algorithmic governance and information equity

Education

• The Ohio State University

  Bachelor of Science - BS, Physics

• Miami University

  Bachelor of Science - BS, Business/Commerce, General

• University of Washington

  Master of Science - MS, Information Management