Selected work
A focused selection of data engineering, data science, and analytics projects — impact first. Filter by discipline, search, or let chance decide.
A complete modern data-warehouse build — ingestion, transformation, dimensional modeling, and BI — that turns raw operational data into analytics-ready tables for downstream teams.
Demonstrates multi-source ingestion and a medallion-style Lakehouse design built for explainability — combining behavioral, environmental, and demand signals.
Scores streaming order data for fraud as it arrives rather than in batch — the streaming-infrastructure skillset data engineering roles look for.
An end-to-end streaming scoring loop — event ingestion, model inference, and synthetic-data generation — on a managed Postgres backend.
Builds an analytics-ready dataset from a massive (millions-of-records) source — large-scale ingestion and normalization into Delta for downstream ML and reporting.
A resilient, incremental ETL pipeline — async I/O plus caching means subsequent runs only fetch missing data, keeping rate-limited API usage efficient.
Demonstrates pipeline orchestration — turning loose scripts into scheduled, dependency-aware, observable workflows, a core data-engineering competency.
Hands-on with the industry-standard orchestrator — scheduling and managing dependencies across multi-step data workflows.
Shows applied deep learning on sequential data (RNN + attention) aimed at saving manual fraud-review labor and preventing losses before they happen.
A complete computer-vision workflow — data prep, custom model training, inference, and tracking — on real video footage.
End-to-end model development on audio data — feature engineering, multilabel training, and export to a deployable format.
The deployment half of the ML lifecycle — shipping a trained model to users with zero serving cost, no cold start, and reproducible preprocessing parity.
Codifies a repeatable, modular data-science workflow — useful signal that work is structured and reproducible rather than ad hoc.
Analyst-facing work — defining KPIs, modeling the data, and designing reports that connect an external signal (weather) to a business lever (pricing).
The analyst loop end to end: explore raw data in Python, build a reusable semantic model, and ship a stakeholder-ready report.