Working as part of the Advanced Data and Analytics Program (ADAP)

EVA is a digital assistant creation platform. It offers innovative and automated customer engagement through assistant over Video telephony call, assistant over Audio telephony call and 3-in-1 (Video/Audio/Text) assistant over conventional channels (Mobile App and Web browser).Enhanced responsiveness of the assistant by building Intent classification and Named Entity Recognition models and merging them using a wrapper API to call other APIs based on given customer tasksAlso designed RAG ( Retrieval Augmented Generation) based question answering pipeline and an evaluation pipeline to get the contextual accuracy of the generated answers for a given question answer pair datasetTools and Technologies:-LangChain, LangSmith, Hugging Face Transformers, Tensorflow, Torch, Python

Modules learnt :-1) Supervised Learning2) Deep Neural Networks3) Dimensionality Reduction4) Unsupervised Learning5) Probabilistic Graphical Models6) Sequential Learning7) Causal Inferencing8) Reinforcement Learning

Worked on medical image analysis and detecting OSCC.The results obtained were promising and published in MDPI Diagnostics journal.