About

I'm drawn to problems in ML-driven AI systems from a systems and architecture perspective, particularly where data pipelines, orchestration, and long-running workflows must behave predictably under real-world constraints: systems with predictable failure modes, tractable architectures, and durable operability. I'm less interested in isolated models or surface-level AI features, and more interested in how these components are composed into systems where engineering discipline matters, and where correctness and robustness are as important as raw capability.

I favor explicit contracts over implicit assumptions, and systems designed to be understood, not just built. I'm drawn to novel complexity when it solves real problems, but I'm skeptical of complexity that exists to be clever. When evaluating trade-offs, I lean toward approaches that make failure visible, keep debugging tractable, and leave room for the next engineer to reason about the system without needing its original author.

Outside of formal roles, I actively experiment with ML and AI frameworks, build hands-on projects, and refine my understanding of how AI systems can be structured, evaluated, and maintained in practice. More broadly, I enjoy building systems in general, especially when the work involves clear problem definition, thoughtful design, and turning abstract ideas into working software.