UDAY KIRAN REDDY KONDREDDY

Worked on luis bot framework to develop a multiple virtual assistants on a single chatbot for a client. Worked on RASA to develop a chatbot for a hospital help desk and also modified clinicalBERT according to the requirement.

Developed ML models to analyze IT ticket conversations and server data, including sentiment analysis, keyword detection, and anomaly detection on server and storage data. Implemented scalable experiment tracking and data processing platforms by deploying MLFlow across multiple servers, establishing a standalone Spark architecture, diskover application to scrape filesystem metadata.

To reduce manual effort, we built a model that analyzes discovery logs and focuses on new alerts via clustering, reducing the human time needed for triage. We also provide error log insights using our in-house LLM models. ServiceNow customers often have numerous workflows and clones, which can consume significant time and resources. To optimize this, we created a model that accurately estimates workflow runtimes. This allows customers to better plan and prioritize workflow execution to maximize efficiency. To help our SRE team focus on high-priority alerts, we developed an LLM model that filters out low-severity alerts and suggests remediations from our knowledge base.

I built and deployed a robust ML pipeline on Databricks that generates personalized recommendations for thousands of merchants. I also developed a user-friendly tool for clients to visualize product recommendations and analyzed out-of-stock revenue data to highlight missed opportunities

Awarded with a 25K cash prize and top team among 80 teams.