A deterministic workflow engine for modern AI applications that validates workflow topology before execution. Built to eliminate invalid states, provide predictable routing, and improve observability through OpenTelemetry, with support for YAML-defined workflows, CLI tooling, and seamless integration with the OpenAI Agents SDK.
▶ View Project
A schema-aware YAML configuration system for ESA Pyxel simulation modes, designed to reduce misconfiguration risk in scientific workflows. Implements guided configuration flows, integrated validation, and contextual error handling to enforce structural correctness and improve usability in complex detector simulation environments.
▶ Visit
A technical portfolio of contributions across backend, frontend, and documentation within the GitLab ecosystem. Focused on validation, data integrity, and preventing invalid system states through safer parsing, input constraints, and improved API clarity in large-scale production systems.
▶ Visit
A full-stack geospatial application integrating a FastAPI backend with the Copernicus Sentinel Hub API for real-time Sentinel-2 imagery, NDVI computation, and statistical aggregation. Orbital mechanics are computed via Skyfield (SGP4 propagation) to contextualize satellite passes. The frontend uses CesiumJS to render an interactive 3D globe with toggleable data overlays. Render hosting the API.
▶ Visit
An exploratory contribution site documenting my journey into OpenTelemetry - the open source standard for distributed observability - alongside the ideas that drew me to it. Frames observability as a second-order cybernetic practice, where engineers, users, and the systems they build all sit inside the same feedback loop. A working notebook for contributions, reflections, and the long-term question of what it means to make complex systems legible to the humans who depend on them.
▶ VisitLong-form technical writing — tutorials, comparisons, and architecture thinking behind the tools. Clear writing is part of how I build.
A step-by-step guide to catching structural problems in an AI workflow — dead ends, invalid transitions, and non-terminating loops — at load time, before they reach production. Uses Python, YAML, and topology validation.
📄 Read on DEV.to →An evenhanded comparison of the two platforms — integrated DevSecOps suite versus composable ecosystem — covering CI/CD, security scanning, project management, and how to choose based on how your team actually works.
📄 Read on DEV.to →Why most system diagrams hide the risk — and how to draw trust boundaries, typed data flow, and data persistence so they're visible. Walks one system through three stages of scale to show how the risk surface grows.
📄 Read on DEV.to →