Machine learning, data engineering, and analytics work. Each project shows its impact and the responsible AI check behind it.
Multi-magnification deep learning on 9,109 tissue images, with Grad-CAM so every prediction shows the tissue behind it.
Impact: Fusing 40X and 200X reached 98.94% recall on malignant cases and 96.38% accuracy.
Responsible AI check: Recall was the priority metric because a missed cancer costs far more than a false alarm.
A recommender for 31M real H&M transactions that optimizes for what you’ll keep wearing, with one dial that trades accuracy for sustainability.
Impact: A LightGBM LambdaRank model reached MAP@12 of 0.0247, about 3x the popularity baseline.
Responsible AI check: Every point on the accuracy-vs-sustainability curve ships with a 95% bootstrap interval.
SARIMA forecasting on 116 years of monsoon data, testing whether forecast-interval breaches can flag floods like 2022.
Impact: SARIMA beat a baseline ARIMA model by 15% on RMSE.
Responsible AI check: A naive historical-average baseline came within 0.03 mm, and the write-up says so.
Scheduled ETL over Denver’s 311 feed: S3 ingestion, PySpark, quality gates, a Postgres star schema, and a live dashboard.
Impact: 69 automated tests and an 80% coverage gate in CI.
Responsible AI check: Quality gates fail the run before bad data ever loads.
A cognitive health app for older adults: five-minute daily brain exercises that start from a 30-day built-in program, then keep going with fresh activities generated by the Claude API.
Impact: A 30-day exercise bank runs fully offline, then the Claude API keeps the program fresh with new daily activities.
Responsible AI check: Claude generates content but never grades the user, and personal data never leaves the device.
Do brands that disclose more actually behave better? 247 brands, two independent indexes, one honest answer.
Impact: Transparency explained only 14% of sustainability performance (r = 0.37).
Responsible AI check: Reported as a weak relationship, with the small overlap (n = 49) stated up front.
Tested an organization’s assumed driver of community outcomes on 50,000+ records across 78 neighborhoods.
Impact: The assumed variable barely correlated. Poverty rate explained most of the pattern, redirecting a $500,000 strategy.
Responsible AI check: Findings reported as associations, not causal effects.
Linked census, H-1B, and patent data to find which skilled worker groups drive patenting in Silicon Valley.
Impact: H-1B approvals were the strongest predictor of patenting, led by STEM-educated immigrant workers.
Responsible AI check: Observational data, so conclusions are stated as associations.
I'm glad to talk about interpretable ML, LLM systems, or a dataset that deserves better.