Marketing Data Scientist building forecasting, experimentation, and causal measurement systems for retail/eCommerce — turning messy behavioral data into decisions.
Previously: Nike • adidas • Intel | Python • SQL • Databricks • Snowflake • Streamlit

🔗 LinkedIn • ✉️ Email • 🧭 Retail Trend Tracker (live)

Note: Everything here uses publicly available and/or synthetic data, designed as reusable templates that can be adapted to real internal datasets.

Repo: https://github.com/jwitcher3/m5-causal-lift
End-to-end incrementality sandbox on M5-style simulated retail data with known ground truth.

• Methods: DiD / Event Study • Synthetic Control (ridge SCM, optional log1p)
• Trust checks: pre-trend checks • diagnostics (pre-fit RMSE, stability CV) • placebo tests (fake treatment dates/windows)
• Includes a Streamlit app + donor weight interpretability
• Run: ./scripts/demo.sh (launches Streamlit locally)

Live: https://retail-trend-tracker.vercel.app/
Repo: https://github.com/jwitcher3/retail-trend-tracker/
Deployed dashboard surfacing retail/sneaker trend signals to quickly see “what’s up / what’s down.”

• Focus: franchise-level trend monitoring and lightweight public dashboards
• Stack: Python • JavaScript • Plotly • Vercel

Repo: https://github.com/jwitcher3/edgar-retail
App: https://edgar-retail-etl-jw.streamlit.app

End-to-end ETL + analytics app that pulls public SEC EDGAR filings (10-K / 10-Q) + XBRL financials for selected retail brands and converts them into clean quarterly tables + an explainable pressure signal.

• Outputs: DuckDB warehouse (bronze/silver/gold) + exportable CSV/MD/PDF/ZIP bundles
• Signal: pressure_index = normalized filing-language pressure + inventory-to-sales context
• Evidence engine: ranks and highlights the best filing excerpts (with SEC links) behind each quarter
• Use cases: earnings prep, competitive monitoring, promo/markdown intensity tracking, planning risk signals

Methods: forecasting • experimentation • causal inference • segmentation • attribution/CLV
Build: dashboards • data apps • pipelines • reusable analytics templates
Stack: Python • SQL • Databricks • Snowflake • JavaScript • Streamlit • Plotly • Azure • DuckDB • Parquet
Dev / Tooling: Git • Make

• Retail, eCommerce, and product analytics projects
• Public dashboards, LLM-enabled insights, and visual storytelling
• Data science education or training resources