Deepayan Banerjee

Built AI-driven plant intelligence platform using LLMs and multi-agent orchestration for real-time insights.

• Implemented RAG pipelines to enhance context-aware responses from plant documentation and historical data.
• Reduced expert intervention time through autonomous AI decision support.
• Developed semantic product search engine with LLMs and vector embeddings for intelligent discovery.

Replaced keyword-based searches with natural language understanding, boosting accuracy and user experience.

• Integrated enterprise product catalog with RAG for timely and accurate responses.
• Architected AI analytics platform automating insight generation from operational and business data.
• Leveraged multi-agent workflows to automate data analysis and streamline narrative report creation.

• Created machine learning pipeline to predict end-of-line test bench times, streamlining car sequencing process.
• Developed classification model for fault prediction in assembly line cars, achieving accuracy of 86%.
• Achieved R² score of 67%, enhancing plant managers' ability to manage test bench efficiency.
• Implemented ETL processes for data engineering, processing, and dashboard deployment using Python and Tableau.
• Developed end-to-end pipeline in Python and Qlik Sense for rear axle assembly analysis, enhancing data accessibility for stakeholders.
• Automated live data connections, eliminating MATLAB license costs and complexities.
• Designed cockpit dashboard visualizing tolerance levels, facilitating identification of parts needing rework.
• Established live data visualization in factory, resulting in significant time and cost savings.

• Upgraded OpenSSL from version 1.0.2f to 1.1.0 to enable support for TLS versions 1.2 and 1.3, improving security compliance.
• Updated TLS security version for modems used in Apple iPhone, enhancing security protocols | Tech Stack: C, GTI Framework, Networking, Jenkins
• Replaced deprecated functions with safer alternatives in Data Protocol, improving code reliability and security.
• Worked on Data Protocols to increase the Condition Coverage by writing the test cases.
• Worked on the Ethernet Module to send the packet from LWIP to the Host side(AP) so that it will be helpful to trace the packets from the host side.