Abhay Sudhakaran
RSS FeedI build data platforms for pharma and life sciences — Databricks, Salesforce Life Sciences Cloud, and everything that connects them.
Senior Solution Architect / Data Engineer working in consulting, with a focus on lakehouse architecture, CRM-to-warehouse integration, and MLOps for regulated environments.
Selected Projects
- Lakehouse for Commercial & Field Analytics
A Databricks lakehouse consolidating commercial, sales, and field-force data for a pharma manufacturer, replacing a patchwork of regional warehouses with a single governed source of truth.
- Databricks
- Unity Catalog
- dbt
- Delta Lake
- Azure
Case study in progress
- Salesforce Life Sciences Cloud ↔ DWH Integration Framework
A bidirectional sync framework between Salesforce Life Sciences Cloud and the enterprise data warehouse, giving field teams and analysts a consistent view of HCP engagement data.
- Salesforce Life Sciences Cloud
- MuleSoft
- Snowflake
- Databricks
Case study in progress
- MLOps Pipeline for Demand & Commercial Forecasting
A validated MLOps pipeline on Databricks for demand-forecasting models, built to satisfy both engineering standards and pharma change-control requirements.
- Databricks
- MLflow
- Feature Store
- Airflow
- GxP
Case study in progress
Recent Writing
The $0/month pipeline: running a production-grade daily system entirely on free tiers
Automated ingestion, a deterministic rule engine, push alerts, and two dashboards — with a monthly infrastructure bill of exactly zero. Here's the stack, the gotchas, and why the constraints made the architecture better.
Unity Catalog as the Governance Layer for Pharma Data Platforms
Why Unity Catalog's fine-grained access model maps well onto pharma's need to segment commercial, medical, and clinical data without standing up separate platforms.
Getting Salesforce Life Sciences Cloud Data Into a Lakehouse Without Fighting the CRM
Patterns for syncing HCP engagement, call plan, and sample data out of Salesforce Life Sciences Cloud without turning the integration into a second system of record.
MLOps Under Change Control: Making GxP and CI/CD Coexist
Notes on running an MLflow-based training and deployment pipeline that also satisfies pharma change-control and model-validation requirements.