Leading the Intuit Assist project with LangChain, integrating documents into AWS Elasticsearch vector stores for semantic searches. Employing GPT-3.5 and GPT-4 models for advanced user response via Retrieval-Augmented Generation on vector stores.

• Worked on improving Intuit's conversation platform for better chatbot systems.

• Worked with managers and designers to gather project necessities and investigated options for Natural Language Understanding (NLU) to work within our system, collaborating with representatives from 16 companies.

• Created a framework to test the performance of NLU engines like AWS Lex, IBM Watson, Google Dialogflow CX, and Microsoft CLU. This framework specifically measured the F-1 scores: accuracy, precision, and recall making sure we chose the best engine.

• Pushed for new features such as Multi-region Redundancy for Lex and worked closely with managers from AWS Engineering.

• Successfully combined Lex with our bot design system for more efficient bot creation, partnered with various teams to implement multi-region redundancy - training bots in various locations to ensure continuous service.

• Partnered with the cross-functional team to strategically brainstorm and formulate a tailored list of APIs for integration with the external service.

●Working on Vmanage, a distributed cloud based multi-tenanted controller which is used to manage the SD-WAN network for enterprise companies.

● Using Neo4J as the graph database and Kafka as the message-queue. Using ElasticSearch to store data which is increasing over time.

● Developed an innovative TestNG framework for which developers can write Functional Tests in XML and can skip Java to improve code coverage.

●Working on Integration of Viptela's SD-WAN technology into Cisco.(https://www.sdxcentral.com/articles/news/cisco-is-ready-to-upgrade-a-million-routers-with-sd-wan-software/2018/08/)