Hey, I'm Abhay!
RSS FeedI build highly scalable data platforms for enterprise clients — Databricks, Microsoft / AWS Cloud, and everything that connects them.
Selected Projects
- Databricks Integration at Scale for Salesforce Life Sciences Cloud
A YAML-configured Databricks pipeline syncing master data into Salesforce Life Sciences Cloud across 9 objects, with zero code changes needed to add a new one.
- Databricks
- PySpark
- Unity Catalog
- Delta Lake
- Salesforce Bulk API 2.0
- Airflow
- Rules-as-Data Quality Framework for Databricks
A reusable data quality validation framework for Spark and Delta Lake: rules live in a config table, checks run in parallel, and every result is logged for audit.
- Databricks
- PySpark
- Delta Lake
- Unity Catalog
- Stock Prediction MLOps Pipeline
An end-to-end MLOps pipeline predicting next-day S&P 500 direction, from feature engineering through automated weekly retraining and Kubernetes deployment.
- Python
- LightGBM
- MLflow
- FastAPI
- Docker
- Kubernetes
- GitHub Actions
Personal project, open source
- Market Sentiment Analyzer
A daily automated system that ingests market data, classifies it through a deterministic rule engine, and pushes threshold-crossing alerts, running entirely on free-tier infrastructure.
- GitHub Actions
- Supabase
- Cloudflare Pages
- Cloudflare Access
- ntfy.sh
Personal project
- Boundary-Scoped dbt Framework for Databricks
A dbt-on-Databricks project that builds CRM-ready source tables through a config-driven medallion pipeline, and deliberately stops at the gold layer instead of owning the sync into Salesforce.
- dbt
- Databricks
- Unity Catalog
- Delta Lake
- Airflow
Recent Writing
Data Quality Rules Shouldn't Need a Code Deploy
A config-driven data quality framework for Databricks. Why rules-as-data beats rules-as-code for business-specific checks, and how it compares to dbt tests and Great Expectations.
How I Built This Site With Astro
A practical walkthrough of the real decisions behind this site: content collections, Tailwind v4 design tokens, static generation, and a few things the tutorials don't cover, like keeping a production site safe while you're still tweaking it.