I bring the data in, build the analytics, and turn it into strategy.
I run ecommerce for a wellness brand, I'm finishing an MBA in data analytics with a finance and accounting specialization, and I'm a sports fanatic at heart. The projects here are how I'm moving toward finance: real cloud infrastructure, real data, and an answer at the end.
leahkatinsights.com · LinkedIn · leah@leahkatinsights.com
| Project | What it does | Built with | |
|---|---|---|---|
| 3 | Lululemon DCF Valuation Pipeline | SEC EDGAR XBRL to bronze / silver / gold, SQL window functions, a 5-year DCF with bear / base / bull scenarios, and a reverse DCF that backs out what the market is pricing in | Python, DuckDB, Parquet, scipy |
| 1 | Stock Market Analytics Lakehouse | Yahoo Finance to S3, an S3-triggered Lambda that summarizes new files through AWS Bedrock, Delta Lake and SQL trend analytics on Databricks | AWS S3, Lambda, Bedrock, Databricks |
| 2 | Customer Lifetime Value Model | 65k synthetic golf-retail transactions to RFM features in SQL, then BG/NBD and Gamma-Gamma models for 12-month revenue per customer, on a Delta pipeline | Databricks, Delta Lake, SQL, lifetimes |
| 4 | Multi-Channel Reorder Forecasting System | Recency-weighted velocity model and a five-state reorder engine that turns Shopify and Square sales history into a daily purchase-order queue | Streamlit, Postgres, Shopify and Square APIs |
| 5 | Hotel RevPAR Performance Readout | STR-style monthly readout for an urban hotel: occupancy, ADR, RevPAR, and MPI / ARI / RGI against a comp set, from 119,000 real bookings, with a one-page GM memo | Databricks, Delta Lake, PySpark, pytest |
Full writeups with charts and the reasoning behind each build: leahkatinsights.com/projects
- I like taking a messy dataset and turning it into actionable strategy.
- I build the pipeline first. Bronze, silver, gold on Databricks when the project needs it. Data brought in, cleaned up, and made presentable.
- A lot of the data worked in SQL. There is one clean row per customer, hotel, or SKU before any model sees it.
- If I can't rerun it and get the same answer, it isn't finished. There are seeds, tests, and a source line for every chart.
- I end with the decision, not the notebook. A memo a GM would actually read, or a list an operator can work through with ease.
Certification in Hotel Industry Analytics (CHIA), CoStar Group · AWS Machine Learning Foundations · AWS Cloud Foundations · Google Data Analytics · IBM Data Analyst · IBM Data Science
- MBA, Spring 2027: Data Analytics Concentration, Finance & Accounting Specialization
- Next up on the portfolio: Vail Resorts Monte Carlo, Yankees cost-per-win, Augusta analytics
- Currently learning and prepping for AWS Solutions Architect