darshan bhor

Built scalable ETL pipelines for bank’s subledger data using AWS Glue (Spark), Apache Airflow, and PySpark to ingest and transform data into Amazon S3 (partitioned Parquet data lake). Enabled low-latency analytics via AWS Athena and Presto, powering daily Tableau dashboards for 500+ stakeholders. Built a LangChain-based RAG pipeline using FAISS for semantic retrieval, enabling large-scale service request categorization and summarization; implemented dual-LLM validation for hallucination detection, reducing manual effort from weeks to 5–7 hours. Orchestrated Airflow data workflows with integrated Data Quality checks and alerting using Pandas & Camelot; implemented parallel multiprocessing in Python to accelerate processing, ensure dashboard accuracy, and reduce weekly operational effort by 20+ hours.

Integrated REST APIs for multiple banks to connect seamlessly with backend systems, enhancing data exchange and process automation. Designed and implemented algorithms for fetching, processing and cleaning data, improving the efficiency and accuracy of a reconciliation application. Created interactive dashboards in Grafana to visualize data usage reports, enabling real-time monitoring and insights.

Designed workflows for electronic circuit assembly circuit development to enable real-time monitoring of saline solution levels. Implemented data collection workflows with real-time sensor integration and applied advanced analytics techniques for efficient measurement solution. Published saline level monitoring project in IJERT, focusing on sensor integration and firmware design.