01
Selected work
Three projects that show the range — machine learning in production, an enterprise data platform, and generative AI applied to real content. Titles link to the full case studies.
01
AI / Machine Learning
Completed
Eliminate data-quality issues across a large document corpus by detecting duplicate records, clustering similar entities, and letting human reviewers teach the model to get better over time.
What it does
- Custom dedup algorithm — TF-IDF + cosine similarity finds likely duplicates and groups them into reviewable clusters.
- FastAPI processing backend handles data processing and model integration at scale.
- Next.js review UI where the team approves or rejects matches to keep data quality moving.
- Feedback-driven retraining turns every human correction into an automated learning loop.
80% of duplicates removed
50K+ documents in the corpus
Self-improving via review feedback
FastAPINext.jsPythonMachine LearningPostgreSQLREST APIs
02
Data Engineering
Completed
A production-grade data engineering platform on Microsoft Fabric, delivering reliable, analytics-ready data for enterprise Power BI reporting.
What it does
- Automated pipelines ingest and transform data from multiple sources on a scheduled basis.
- Bronze → Silver → Gold layering follows Medallion Architecture best practices end to end.
- Git-based version control inside Fabric keeps pipelines and workspaces manageable.
- Warehouse models & Power BI dashboards turn the Gold layer into business insight.
Enterprise Power BI reporting
Scheduled multi-source ETL
Governed via Git in Fabric
Microsoft FabricPySparkSQLPythonLakehousePower BI
03
Web Application · GenAI
Completed
Modernize traditional books with AI — users upload, write, and update books while generative AI restyles content, designs covers, and breathes new life into old material.
What it does
- Django platform for uploading, authoring, and editing books with a TinyMCE editor.
- Legacy migration brought old books in through Azure Blob Storage and the Watcher API.
- Generative AI — Fooocus and OpenAI APIs enhance content and generate covers; LangChain + ChromaDB vectorize it for retrieval.
- Multi-agent orchestration generates complete AI-authored books from uploads or PDFs, on an Azure VM built for scale.
65% faster content production
End-to-end AI book generation
Legacy library migrated
DjangoPostgreSQLLangChainOpenAIChromaDBAzureTinyMCEWebSocket