{"id":4462,"date":"2026-04-09T12:11:53","date_gmt":"2026-04-09T16:11:53","guid":{"rendered":"https:\/\/sites.miamioh.edu\/cse-senior-design-expo\/?p=4462"},"modified":"2026-04-21T15:31:13","modified_gmt":"2026-04-21T19:31:13","slug":"ai-augmented-automl-toolchain","status":"publish","type":"post","link":"https:\/\/sites.miamioh.edu\/cse-senior-design-expo\/2026\/04\/ai-augmented-automl-toolchain\/","title":{"rendered":"Agentic AutoML Platform"},"content":{"rendered":"\n
\n\n \n \n \n \n \n \n
Vision<\/td>\n Machine learning has the power to transform every discipline. However, a global shortage of 4M+ AI specialists, compounded by an 80% enterprise AI project failure rate, means that most organizations and individuals who would benefit from ML remain locked out. Today’s tools perpetuate this divide: fully automated platforms deliver opaque results that users cannot explain, trust, or refine, while manual frameworks demand years of specialized training. Our vision is to eliminate this false choice by making the full power of machine learning accessible to anyone with domain expertise and curiosity. We see a future where building a reliable and transparent ML model is a guided experience that deepens the user’s understanding with every step taken.<\/td>\n <\/tr>\n\n
Mission<\/td>\n We implement an agentic LLM-driven ML platform that orchestrates the complete ML lifecycle from data ingestion and exploratory analysis through processing, feature engineering, model training, experimentation, and deployment. An LLM provided with domain-specific tools and deep contextual understanding serves as a collaborator throughout the workflow. It can propose transformations, build interactive notebooks, answer questions through NL queries, and explain its reasoning at every decision point. We prioritize reliability and safety through sandboxed execution and human-in-the-loop approval to preserve user control. The result is a platform that handles the complexity users shouldn’t have to: data cleaning, feature construction, hyperparameter optimization. Unlike conventional AutoML, users retain fine-grained control and every interaction deepens their understanding of their own data.<\/td>\n <\/tr>\n\n
Team Members<\/td>\n Shree Chaturvedi (Product Architect), Ayush Yadav (Lead Engineer)<\/td>\n <\/tr>\n \n
Client<\/td>\n Shree Chaturvedi<\/td>\n <\/tr>\n\n
Site<\/td>\n \nagentic-automl.vercel.app\/<\/a><\/td>\n <\/tr>\n <\/table>\n\n