Shamhith. Kamasani

- Mentored 100+ students on developing, deploying, and monitoring end-to-end ML / LLM projects, leveraging tools like TensorFlow, Docker, Kubeflow, Airflow, MLFlow, Git, Vertex AI, and GCP, while fostering best practices in MLOps, including LLMOps workflows with PromptFlow and BedRock - Organized an MLOps Expo showcasing projects from 25+ teams, featuring real-world applications and automated workflows, while contributing to a public MLOps repo with comprehensive lab materials and tutorials to support scalable ML systems

- Fine-tuned a Llama-2 70B model and implemented RAG infrastructure for a PoC on 500GB of enterprise equipment data, achieving a 15% improvement in prediction accuracy for equipment diagnostics. Utilized shapelets and Visibility Graph models to improve time series classification accuracy by 10% - Collaborated with the AWS team to architect a scalable pipeline that ingests 10GB+ daily sensor data and unstructured textual data from 7+ sources, enabling real-time insights for on-premise LLM deployments - Delivered two internal research memos and a 30-page report, detailing the project phases, including transformers, prompt engineering, RAG, fine-tuning processes, and their impact on LLM response quality

- Led instruction on foundational concepts of supervised and unsupervised learning, providing hands-on Python exercises focused on data wrangling, EDA, and modeling with real-world datasets. - Assist students with their ML capstone projects, offering guidance on modeling strategies, and optimization techniques; contributed to articles on Convex Optimization and data sampling methods on professor’s website

- Conducted data wrangling and exploratory data analysis (EDA) using Python, SQL, and Pandas, involving stakeholders; Built a Multi-Layer Perceptron (MLP) architecture, achieving 90.5% accuracy in predicting structural failures - Optimized data pipeline using PyTorch-based deep-learning workflows, integrating CNNs for feature extraction, boosting data throughput by 30%, and streamlining ETL workflows

Coursework: -Probability and statistics -Deterministic Operations Research -Statistical Learning for Engineering -Data Management for Analytics -Computer Visualization -Machine Learning Operations (MLOps) -Algorithms -Generative AI

Annual Fest: NIRVANA'22 (Student Convener) Lead Sponsorship wing for the fest. Student council: Class Representative(2018-2022) Sports: Basketball team(2021-2022)