Fullstack Developer, Prompthunt
Mar, 2024 - Jun, 20251 year 3 months
Sole full-stack engineer for an AI-powered image-generation SaaS that produced 720 000 images in its first year. Designed and built the complete, serverless solution—covering model fine-tuning, job orchestration, real-time updates, analytics, and user onboarding.
Processed 720 000 images in 12 months with an average 4 s end-to-end latency via a Redis/Upstash queue and asynchronous workers.
Architected and deployed a cost-efficient GPU pipeline on Modal Labs and fal.ai, cutting cost-per-image by 50 % while supporting effectively unlimited queued throughput.
Implemented on-demand fine-tuning for multiple Flux models by automatically provisioning custom GPU instances.
Integrated webhooks and Pusher to deliver real-time status updates and notifications to users.
Built the entire frontend (Next.js on Vercel) plus admin analytics dashboards, onboarding flows, and post-processing utilities.