somanadh maganti

• Leading a RAG based NAVI-Sales project to enable Generative AI capabilities using Microsoft Azure Open-AI GPT-3.5, GPT-4.0 for creation of client proposals for sales community. • Implemented NAVI services utilizing the Vector databases Elastic, FAISS along with AI frameworks Haystack, Lang chain, Llama Index. • Implemented various NLP preprocessing methodologies NER, Summarization, Topic modeling, Tokenization, Clustering etc. for cleaning and making data into structural format. • Successfully implemented NAVI-Employee project by utilizing open source models like Falcon, Llama, Bloom, etc. and fine-tuning the open-source models using PyTorch framework and along with RAI implementation. • Enhanced the quality of solutions from by implementing Prompt engineering, data cleaning techniques and evaluated using various AI metrics like F1-score ,perplexity score, Blue score, Rouge score, and Cosine similarity. • Implemented the End to End application using Python web application frameworks like Flask, Fast API and created CI/CD pipelines for deployment of ML models and worked with Docker , kubernetes concepts and various DevOps tools Rancher , Kibana , Instana

• Successfully Implemented NLP and data science based project INFY-ME recommendation engine for widgets recommendations to 3 Lakh Infosys employees. • Quantified and enhanced the INFYME solution implementing various ML algorithms including classification, and regression models along with data analytical techniques. • Created ETL data pipelines for transformation of thousands of documents and Share Point portal integration which enhanced the efficiency of recommendation engine. • Worked on various data preprocessing and cleaning techniques like Outlier detection and Feature Engineering techniques on Infosys article, employee data. • Contributed end-to-end development and timely delivery of Recommendation Engine solution for NEOM and other clients which increased the business revenue.