Cary Fehler

I’m a senior software engineer architecting cloud-native generative AI platforms and full-stack experiences that ship reliably at scale. My work blends system design, model integration, and product-minded engineering to turn ambiguous requirements into production outcomes.

At Adobe (04/2022–Present), I architected and scaled Adobe Firefly, a cloud-native generative AI platform supporting 16B+ content generations. I designed end-to-end multimodal systems for text/image/video generation and collaborative editing (Firefly Boards), built distributed backend pipelines (Node.js, Python), and reduced P95 latency by ~40% through batching, caching (Redis), and optimized request flows.

I also drive scalable infrastructure and deployment practices—building and operating GPU-backed workloads on AWS and Kubernetes with autoscaling and workload scheduling to support millions of requests/month with 99.9% uptime. I integrate LLMs and RAG-style pipelines into production (including partner model integrations and custom fine-tuning), reducing generation latency by ~25% while enabling enterprise-scale brand-safe AI content generation.

Earlier, I helped Stripe Atlas (06/2020–04/2022) enable 100K+ founders across 140+ countries to form U.S. companies, open bank accounts, and start payments in ~2 days. I implemented distributed onboarding workflows (Ruby on Rails / Node.js, Kafka messaging), improved end-to-end latency by ~40%, and achieved 99.9% uptime while maintaining MTTR < 1 hour through CI/CD, testing automation, and operational reliability practices.