Ahmed Katary

Optimized and tuned a custom resource planning system for forecasts in collaboration with data scientists to support procurement of 2.5M chips for production workloads including Google Maps. Built a SQL + Java forecasting pipeline processing ~100 GiB every 4 hours for a four-engineer team responsible for the pipeline and stockout-risk modeling system.

Architected scalable backends using Node.js, Django, and GraphQL. Built Aria, an AI-powered sound management platform using React TS deployed to 50+ stores across the US and Canada, including an API-based backend to monitor live music metadata and an endpoint to recommend songs using mood, energy, and k-means clustering. Built Evolv Academy endpoints for managing users, content, and submissions using Django, GraphQL, SQL, Firebase auth, and Celery for scheduled emails, increasing student enrollment by 50%. Recruited and managed a team of 2 software engineers and led customer lifecycle delivery across 5 full custom projects including specs, hi-fi design deliverables, quotes, and live testing setups; led 20+ client intake processes and converted 5.

Co-authored Style2Fab: Functionality-Aware Segmentation for Fabricating Personalized 3D Models with Generative AI paper; responsible for technical implementation of stylization nerf-based PyTorch segmentation algorithm and landed a $140,000 grant. Optimized spectral clustering for mesh segmentation by converting mesh into a collapsing face graph enabling ~12s processing for 100K+ surfaces (6x speedup).