Kavana Padaki

I am a data scientist and application developer with two years of software-engineering experience at Cognizant, focused on designing, training, and deploying end-to-end ML solutions in production.

I build production-ready models and applications using Python, TensorFlow, and cloud MLOps tools, and I specialize in NLP and large language models using LangChain and the OpenAI API.

My projects include Streamlit apps for model evaluation and sentiment analysis, a DistilBERT fine-tune achieving 92% validation accuracy, and production deployments optimized for low latency. I also delivered analytics-driven improvements such as a 15% uplift in marketing effectiveness from demand-forecasting work.

At Cognizant I automated a resume-extraction pipeline (reducing processing time by 35%), re-architected UI/backend logic, and introduced SSDLC practices that cut production defects by 10%, demonstrating a pragmatic focus on impact and reliability.