Applied AI Systems, Technical Enablement & ML Pipelines
Autoencoder computer vision pipeline (ORNL publication), Gemini full-stack agent, and AI-literacy curriculum for 500+ researchers at Perplexity AI.
Project Overview
Co-developed machine learning pipelines automating Transmission Electron Microscopy analysis with a four-university team (Penn State, NYU, Purdue, UTK) using autoencoders, published in Machine Learning: Science and Technology (DOI: 10.1088/2632-2153/ae1f5d). Shipped a full-stack Node/Express conversational agent on the Google Gemini API incorporating dynamic elicitation design and intent-conditioned prompting. Built and delivered a technical AI-literacy curriculum to 500+ researchers, faculty, and staff at Perplexity AI, with department-adopted responsible-use guidelines and AI-assisted workflows aggregating verified briefs.
My Role
Machine Learning Engineer & Campus Strategist. Developed computer vision autoencoders, built full-stack conversational web apps, and delivered enterprise AI curricula.
Tools / Stack
Project Details
Implementation Roadmap
Development Process.
Co-develop convolutional autoencoders and data augmentation pipelines at ORNL to automate nanoparticle TEM image analysis.
Evaluate geometry correlations and co-author peer-reviewed publication in Machine Learning: Science and Technology.
Engineer a conversational agent on Google Gemini API that conditions prompts on clarified user intent.
Deploy full-stack Node.js/Express web app with session state and responsive browser UI.
Build and deliver AI-literacy workshops for 500+ researchers, postdocs, and faculty, drafting department responsible-use standards.
Case Study Analysis
Challenges & Outcomes.
Creating ML models and AI workflows that combine rigorous computational performance with transparent, accessible adoption for researchers.
Built modular Python/PyTorch computer vision architectures and structured hands-on educational curricula for structured prompting and source verification.
Published peer-reviewed ML paper with ORNL collaborators, deployed a full-stack Gemini agent, and trained 500+ campus researchers.