Jhilam Ray

• Designed and implemented end-to-end data pipelines/ETL, including requirement gathering, design, development, and implementation.
• Designed and delivered a robust Data Governance Tool from scratch, featuring data cataloging, lineage tracking, and workflow orchestration, streamlining management of data assets across the team.

https://wiki.corp.adobe.com/display/AcomDE/A.com+%7C+Data+Governance+Tool

• Worked on end to end an AI driven chatbot for an A/B testing dashboard using the GenAI frameworks on Databricks.

https://wiki.corp.adobe.com/pages/viewpage.action?pageId=3613050354

• Generative AI (GenAI) with practical experience in deploying real-world business solutions and leveraging its potential to enhance AI-driven decision-making.
• OneAI is a unified, MCP-native agentic platform that automates analytics, collaboration, and reporting with modular, horizontally scalable components.
  It implements zero-trust security, rich observability, and interoperable connectors to orchestrate AI workflows across frontend, backend, and enterprise systems.
• Automated business summary generation (RTB) using AI-driven frameworks on Databricks, optimizing reporting of weekly business metrics for leadership.
• Data Load from databricks to Adobe experience platform. Query service in AEP.

• Capabilities to detect mismatches/errors in label images of Hard Disk for barcode/text/logo items. All labels being printed can be checked for quality and any disparity in quality can be reported.
• A master library for the Master image level reference as well as FMT and MDL files and the same is being used based on the selection from UI dropdown to compare with Test image level. False detections/over sensitivity being noticed for text comparison of Master vs Test image.
• As this system depends on tesseract for extracting text contents from an image and it fails to detect few texts completely/partially due to small font, the system marks few of the texts as missing though those are present in test image.
• Developed pipeline and automate ML pipeline. Dynamic Model Generation, Evaluation based on ROC(AUC), Comparison based on ROC(AUC), Model hot Swapping, Start inference. Every week Model testing/monitoring based on ROC(internal monitoring).

• Generic big data load framework to easily onboard any kind of fact data to Hadoop. This framework will utilize the generic metadata system to build the load scripts to load the data and then use generic keying components like MDM or ETL Dim to key and finally load the keyed data to Hadoop system. This process will be able to load any fact data with minimal metadata setup. Once the data is landed in Hadoop.

• Commercial Cards Analytics

• Sony ETL developer

• JCPenny

• E-Sushrat