At Irastrive Technologies Private Limited, I integrate OCR-based extraction into Azure data pipelines, turning unstructured documents into structured, validated records stored in MongoDB and ADLS Gen2. This reduced manual effort by about 30%.
I also designed and integrated LLM prompt workflows into production data processing, and built a rubric-driven engine to score AI-generated outputs against defined criteria.
At HCLTech, I designed and maintained 17+ production ETL pipelines using Python, SQL, PySpark, Azure Data Factory, Databricks, and ADLS. I implemented medallion architecture with Delta Lake and star schema models to organize data for reporting.
I implemented incremental loads and SCD Type 1/2 logic that reduced load time by 75% compared to full loads. I also built automated data validation and test suites integrated into CI/CD pipelines, reducing data quality issues by about 20%.
