I build reliable, data‑driven systems and human‑centered AI. Recently: vector DB + LLM features for SASVA, imaging/health projects, and clean backend services on AWS.
I collaborate across product, data, and design to ship dependable software. My toolkit spans Python, JS/TS, React, Node, SQL, Docker, and AWS. I care about clarity, performance, and shipping features that actually help people. Lately I’m focusing on healthcare + NeuroAI projects.
Accessible ML/NeuroAI explainers, TikTok/YouTube series, and hands-on demos.
I design bite-size curricula, live code walkthroughs, and visual explainers (activation functions, attention, EEG features). The goal is clarity without losing rigor.
EEG → features → real-time insights for state detection and biofeedback.
Stream Muse2 signals, compute band-power (δ/θ/α/β/γ), and surface uncertainty-aware inferences. Built with Python, PyTorch, and a lightweight web UI.
NIH ChestX-ray14 multi-label classifier with clean data pipeline & evals.
Custom PyTorch dataset, 224×224 preprocessing, threshold-tuned sigmoid outputs, AUROC per label, and error analysis. Roadmap: Grad-CAM + uncertainty.
Mini‑games for 2–4 year olds. Tracks scores per child and surfaces insights for parents.
Find work spots with Wi‑Fi / power near you. GitHub login, location search, suggestions.
Anonymous student chat with RA + Admin broadcast boards. Includes invite/auth code flow.
Quick search UI for trending and themed GIFs via API.
Casual number‑matching web game. Built for practice + fun.
Classic RPG mechanics in a lightweight browser experience.
Interested in collaborating or hiring? Reach out on LinkedIn.