Spearheading state-of-the-art generation of relayable synthetic crack-segmented images and unbiased device classification to maximize precision of e-waste processing.• Built and trained contrastive learning models to classify raw discarded mobile devices by model type for internal component sortation• Researching image augmentation overlay pipeline to transparently scale overlays while preserving aspect ratios• Optimized ML algorithm to autonomously remove device orientation and… Show more Spearheading state-of-the-art generation of relayable synthetic crack-segmented images and unbiased device classification to maximize precision of e-waste processing.• Built and trained contrastive learning models to classify raw discarded mobile devices by model type for internal component sortation• Researching image augmentation overlay pipeline to transparently scale overlays while preserving aspect ratios• Optimized ML algorithm to autonomously remove device orientation and color bias with photorealistic image production, improving confidence by 39% Show less

Led implementation of ML models trained to auto-fill public offerings contracts based on cURL and bulk SEC EDGAR data. • Implemented data filtering and sorting algorithm to parse bulk EDGAR Form 1-A and S-3 filings to develop autonomous financial filing system• Conducted research (paper) and organized security non-registration strategy via SPV for blockchain regulatory analysis

Assisted in pedagogical research surrounding classification by CMU coding scheme of Docuscope-based student writings with the CMU Data Science Department TeachStat Lab.• Extracted BERT embeddings from past student projects with bi-encoders and word embeddings to build SBERT model• Classified sample sentences with k-NN node, estimating with cosine similarity and random forest

Collaboration with the technology team to optimize company performance in both its pre- and post-seed phases.• Developed algorithm to autonomously parse federal grants by relevance and applicability• Creating best dashboard UI to enhance client experience accessing mobile charging hub reports• Upgraded website to highlight company’s revolutionization of electronic vehicle infrastructure industry

Created and managed social media content for mentorship project focused on developing advanced STEAM curriculum for thousands of underprivileged K-12 students.• Taught self-devised coding and other STEM curricula at welfare institutions across the DMV• Produced animated videos in 3D & educational course for remote student engagement during the Covid-19 pandemic• Organized dozens of summer enrichment events for DMV-area family shelters and supply collection drives

Communicated with hundreds of homeowners, HOAs, and brokers to curate niche market trends and demographical data.• Designed interior design simulations using Photoshop and Procreate to advise open house stagings• Organized dozens of coverage events and workshop models- analyzing client feedback to boost representation rates• Ran agents’ real estate marketplaces, social media accounts, and brand websites