Yamini Sri Vandana Penamakuru

I worked in Agile Methodology for below projects

• Conducted research on chatbots such as Amazon Lex, Dialogflow, and Merck self-service chatbot.
• Researched NLP techniques for intent detection and named entity recognition in chatbot systems.
• Initiated AI-driven pilot to evaluate semi-automated compendium review using similarity analysis and entity extraction.
• Created an algorithm for comparing text and highlighting differences across files.
• Optimized retrieval processes by creating an automated solution for API data extraction.
• Created Python test scripts to validate data pipeline integrity.
• Designed a data-driven user interface with Streamlit for dynamic visual representation.
• Developed and executed business capability strategy aligned with overall domain goals.
• Revamped application framework to shift focus from disease-centric to product and drug name-centric.
• Implemented Coveo Intelligent Search for efficient document retrieval solutions.
• General Pre-Review Tool (GRT): The project aims to use an enhanced AI approach in a tool solution that aims to increase efficiency, save time and costs, reduce manual screening efforts and reduce errors, thus increasing compliance and innovation in the departments
• Optimized tokenizer for more accurate keyword combination handling.
• Processed text by utilizing OCR technology from PDFs.
• Examined several Hugging Face NER models, including NER-Chemical-Bionlp, Biomedical-NER, HunFlair NLP, and BERT.
• Worked on the Generative AI summarization of insights system (GENESIS) to generate the summarization of insights, experimented with the LexRank algorithm, and developed AWS Lambda functions, Step Functions, and Terraform templates.
• Engineered synthetic data generation using Generative AI for clinical trial forms.

• Detection of low-head river dams using deep Learning (Project under Federal Government): Worked closely with Polis center to develop a deep learning model that detects low-head dams from Geo Spatial images provided
• Engineered a CNN model leveraging Pytorch with an 84% accuracy rate.
• Early detection of low head dams helps in reducing the risk of accidents for kayakers
• Collaborated with Polis Center to obtain geospatial images of Indiana counties for classification.
• Collected, analyzed, and labeled raw datasets for model development using ArcGIS.
• The project further aims to classify building types using the generated labels.

• Resolved incidents in employee and job database using Core HR module.
• Experienced and skilled in SQL, managing Peoplesoft modules like Enterprise Learning Management, Recruiting, Retrofitting
• Skilled in writing, testing, and deploying code into production efficiently.
• Leveraged Oracle HCM Cloud to handle employee information and track absences.