Architecture

Python

PuppyDog.io — Eleven Services, One GPU, 80% Leaner

How I split a tightly coupled media pipeline into eleven independent, queue-driven services, confined the GPU to the one workload that needs it, and cut monthly infrastructure cost by about 80%.

Node.js

PuppyDog.io — Personalized Demo Infrastructure at Scale

How I built a mass personalization engine, migrated AI compute from AWS to GCP saving 36% on infra costs, and took backend test coverage from 0 to 75% as the sole backend engineer.