Hire Sushmitha T. - Chatbot Developer

Built a RAG (Retrieval-Augmented Generation) pipeline using LangChain and Qdrant vector database to answer candidate FAQs, improving response accuracy and contextual relevance. Developing an Automated Candidate Screening Bot using LangGraph, designed to streamline recruiter workflows and reduce time-to-hire through intelligent candidate shortlisting. Built an Intentless Bot using Rasa architecture, enabling contextual conversation handling without predefined intents, improving scalability and flexibility of chatbot flows.

Working in the AI BOTs team, to build end-end Chatbots for WhatsApp, SMS and multi-lingual (English, Dutch etc.) career sites, for 320 companies with 155k interactions & 60k active users daily. Built the bots using the Rasa framework in 3 steps. Firstly, intent classification & entity extraction – Did pipeline selection, built NLP models with an accuracy of 93% & 85% resp., using selective features like, DistilBERT sentence embeddings, lexical features, regex etc. and tuned the DIET model. And then, dialogue flow – Tuned deep learning models like, TED (Transformer Embedding Dialogue), and used policies like, mapping, memorization etc., to successfully predict 25 actions, based on context, intent & slots, using ensembling methods. Lastly, action execution – Generated bot responses on predicted actions & integrated with other teams upon dependencies. For WhatsApp and SMS bots, used Twilio cloud communication platform for user integration. Working as the data team lead to improve the bots – by increasing the selective features, upon analyzing Sankey diagrams, bot metrics like leads, apply clicks, lead-apply ratio etc., and by increasing the accuracies of the models, upon analyzing insights on annotations of 1.5k data points daily, with Rasa X tool. Automated the offline model training and versioning. Career Site bot: https://helponebillion.com/ WhatsApp & SMS bots: Type Discover Jobs at +1-833-721-1245 & 39522 resp.