I build and ship mobile apps end to end, and run independent machine-learning research on open physics and space data.
Nine apps live on the App Store; research that reproduces and discovers structure in real datasets from CERN, Gaia, SDSS and TESS.
Photo-based macro/micronutrient analysis, barcode and menu scanning, five languages.
CV scoring, ATS-friendly rewriting, interview-question generation, in-app purchases.
Screen-time AI coach built on five Screen Time extensions.
Fully offline wellbeing app: habits, mood, breathing, Pomodoro, four home-screen widgets.
Tap-tempo metronome fusing accelerometer and microphone, with background playback.
Guitar-practice and music apps: live chord/tone scoring, chromatic tuner, ear training.
Discovery-oriented ML on open space data: solar radio-burst CNN (AUC ≈ 0.83), NGC 2516 recovery in Gaia DR3, SDSS spectral anomalies, TESS searches. 33-page report.
61.5M real 2012 CMS collisions from the CERN Open Data Portal; recovers eight resonances (J/ψ, Υ triplet, Z) to under 1% mass accuracy.
When does a GNN beat a permutation-invariant baseline? Deep Sets vs dynamic k-NN ParticleNet on JetNet.
E(3)-equivariant GNNs for protein–ligand binding-affinity prediction — geometric deep learning respecting 3D symmetry.
Live scale/note overlay onto a real guitar neck through the camera — homography fret tracking, on-device detection.
Self-hosted AI workspace.
Interactive laboratory for cellular automata and agent-based models.
A desktop graphical interface for the yt-dlp downloader.