Max Hehera

Built Platelog, an AI calorie tracking app for iPhone that estimates nutrition from a food photo, with 300k+ users and a 4.8 App Store rating from 4,000+ reviews, working across the whole product from the SwiftUI client to the backend, payments and growth systems.

Rebuilt the photo logging flow, cutting the time from shot to result to under 3 seconds, and shipped barcode scanning, editable porti

Built a custom computer-vision pipeline from scratch for in-app document and check capture (automatic occlusion removal and page dewarping), cutting unusable scans by around 60% and removing the manual review step the product relied on before.

Trained, evaluated and deployed the models end to end, from dataset to inference, reaching around 95% extraction accuracy on live production traffic.

Own

Built real-time trading systems for a crypto exchange used by 500k+ traders: WebSocket price feeds, order books and live charts across 300+ pairs, holding update latency under 150ms during live market load.

Optimized rendering and state for high-frequency market data, cutting chart render time by around 40% under peak load.

Implemented secure wallet and KYC flows with WalletConnect and on-chain

Built scheduling, speaker profiles and real-time attendee features for an enterprise event platform serving 50+ corporate clients.

Created a reusable component library and integrated REST APIs across content, media and users.

Raised automated test coverage from around 30% to 70%, reducing regression bugs.