Panagiotis Panageas

Designed and shipped backend services for a customer renewal risk platform combining account data, usage signals, support metrics, and billing status into unified health scores. Implemented a multi-tenant architecture with Stripe billing integration and role-based access control (RBAC) plus API key management for enterprise customers. Built background job pipelines for data aggregation, batch email delivery, and scheduled renewal risk recalculation using async workers. Developed AI-assisted recommendations using OpenAI APIs and designed prompt engineering workflows with streaming responses for real-time decision support.

Shipped and monetized a commercial desktop automation platform with 50+ active paying users. Built a stateful workflow engine with profile-based triggering, robust state machine transitions, and production diagnostics for real-world debugging. Engineered a GPU-accelerated computer vision pipeline (PyTorch + CUDA) to achieve sub-100ms latency for real-time visual automation. Designed an async event-driven architecture using a Redis-backed queue for reliable concurrent task execution and graceful degradation under load. Delivered a full-stack desktop application (Python + PySide6/Qt) with automated update distribution, plus custom crash reporting and telemetry for production observability. Owned feature roadmap, customer support, performance optimization, bug triage, and long-term maintenance with zero downtime.

Designed an end-to-end AI listing generation system integrating OpenAI APIs with custom prompt engineering for real estate-specific property descriptions. Built lead management workflows including inbox organization, CRM integration, and agent analytics dashboards for deal pipeline visibility. Implemented subscription billing with Stripe, automated renewal reminders, and churn analytics while managing recurring revenue operations. Set up CI/CD pipelines with GitHub Actions, containerized services with Docker, and automated testing with Playwright to ensure reliability under production load.