Tirupathi Rayudu Kommineni

Tirupathi Rayudu Kommineni

Master of Science in Data Science

Regis University

Expected Graduation: Spring 2026

About Me

I am a Generative AI Engineer, Machine Learning Engineer & Data Scientist with over seven years of experience designing scalable, production-grade AI solutions. I am currently maintaining a 4.0 GPA while pursuing my Master's in Data Science at Regis University. My expertise lies in bridging robust Data Engineering foundations with state-of-the-art Generative AI and advanced ML modeling. I am deeply specialized in architecting autonomous AI Agents, Multi-Agent orchestration, and complex Enterprise RAG pipelines. I am highly skilled in MLOps, LLM optimization, and cloud-native AI infrastructure to build, deploy, and scale fault-tolerant, reasoning-based AI systems. My primary goal is to automate complex enterprise workflows and drive measurable business impact.

Technical Skills

Programming Languages

Python, SQL, R, Bash/Shell

Tools & Frameworks

Docker, Kubernetes, DVC (Data Version Control), MLflow, GitHub Actions, Git, Elasticsearch

Machine Learning

PyTorch, TensorFlow, Scikit-Learn, XGBoost, Hugging Face (Transformers), LangChain, LangGraph, OpenCV

Web Development

Flask, FastAPI, Streamlit, Gradio

Databases

MySQL, SQL Server, PostgreSQL, MongoDB, Oracle, Qdrant, ChromaDB

Practicum Projects

MSDS 692

Na

Na

Regis University | MSDS 692 | Spring 2025
MSDS 696

A Study of Iterated Prisoners Dilemma using Multi AI Agents​ ​

Large Language Models Multi-Agent Systems Game Theory Data Analysis Natural Language Processing

This project investigates whether AI models learn to cooperate or act selfishly when playing a three-player Iterated Prisoner's Dilemma. Across 79 experiments, open-source models (Llama3, Gemma2, Mistral) were tested under varying parameters like temperature and history window. SHAP analysis revealed that the specific combination of models drives 82.5% of cooperative behavior, overwhelmingly outweighing game settings. For instance, three Gemma2 agents achieved perfect cooperation, while mixed groups saw cooperation drop to as low as 38.2%. Ultimately, model selection determines multi-agent teamwork success far more than any tunable system parameter.

Regis University | MSDS 696 | Spring 2026

Get In Touch

Let's Connect

Feel free to reach out for collaboration opportunities, questions about my projects, or just to connect!

Send a Message

I'm always open to discussing new projects, creative ideas, or opportunities to be part of your vision.

Email Me Download CV