Vyshnavi V

Roles and Responsibilities - AT&T

• Responsible for creating head intents with related training phrases, having accurate entities annotated, and defining queues in the dispatch table of Genesys Designer for agent routing.
• Implemented hybrid steering use cases with GenAI using playbooks and Generators, coupled with traditional NLU, following conversational design.
• Experimented with Gemini Flash and Gemini Pro models to fetch the best results for agent responses.
• Implemented Infobot that retrieves responses by indexing web pages as a data source.
• Solid understanding of the release management process (pipeline promotion) and deployment of virtual agents into production using Google Cloud Functions triggers.
• Analyzed Google's advanced agent modeling (AAM) generated data to ensure verbiage is concise and consistent with business rules.
• Solid understanding of CCAI best practices and design principles, giving a positive persona to the bot.
• Create intents, entities, route groups, state handlers, APIs, as required, and configure transitions to flows/pages with defined routes.
• Proficient in conversational AI best practices and design principles to create bots with a positive persona.
• Architect new Dialogflow user experience on Lucid chart, including error handling and defining expected formats for API/Webhook parameters.
• Developed SMS flow that can send text messages with a template created from Rebrandly for content requested by the user to drive containment rate up.
• Optimizing chat/voice bots, collaborating with data engineers to fine-tune NLU model, and always aiming to drive for targeted containment rate.
• Extensively used Lucidchart and Miro board to design UX or conversational flows and present them in COE panel reviews.
• Meet with feature lead to discuss Privacy Impact Assessment, if required.
• Merge Apigee to Dialogflow user experience with webhook integration.
• Being responsible for QA and UAT test case review process and signoff on new feature in development until it goes into production.
• Optimizing current bot performance by identifying patterns based on flow-level outcome, leveraging data analysis insights to make the bot efficient.
• Worked on challenges around different areas of NLU Model Tuning, Looping, Speech to Text issues, and conversational Design.
• Proficient with Python and JavaScript programming languages.

Roles and Responsibilities – Lowe's Retail Support Chatbot:

• Design engaging conversational interfaces for Contact Center AI, drive virtual agent creation projects with defined scope and complexity.
• Developing chat bots and voice bots (preferably with Dialogflow).
• Used JavaScript, Dialogflow ES, and integrations with Cisco telephony gateway.
• Also, implemented DF connector integration with VA to record transcription of live agent conversations and provide the recordings for storage within Microsoft Dynamics Application.
• Developed authentication process using up to 10 identity validation questions specified by the Unemployment Insurance Division.
• Used REST APIs to pull predetermined data points from UI Claims Status Summary dataset for applicable use case.
• Performed UI/UX development, chat widget customization.
• Hosted 2 virtual training sessions for administrative users to review solution capabilities and critical processes.

• Research on intent detection and named entity recognition approaches for virtual agent.
• Enhancing workflows on both voice and text-based chatbots with help of dashboard analytics.
• Reviewed and improved conversation flows and chatbot implementations (billing, appointment booking) to improve containment rate by 20%, as part of optimization.
• Improved Named Entity Recognition by mapping regex patterns.
• Worked on keyword extraction for annotating entities in training phrases.
• Altering flows or use cases in virtual agent depending on demographic of users.
• Conducted workshops and tutorials on chatbot development (DialogFlow).
• Solid understanding of NLP and machine learning concepts.