AI engineer building agentic AI and machine learning products
I work across engineering, research, and product thinking to ship LLM systems, applied machine learning, and user-facing AI tools with practical business value.

I am an AI engineer focused on turning ambitious AI ideas into real products. My work spans machine learning research/applications, agentic systems, LLM applications, for experiences that need to be both technically strong and genuinely useful.
I particularly interested where engineering, AI, and product strategy meet. That usually means building systems that are not only intelligent, but also safe, trustworthy, and reliable.
This portfolio collects the projects, experiments, and writing that shape how I think about modern AI products.
Open-source orchestration framework for AI agent development, combining marketplace-style tooling, async skill installation, and a structured multi-phase workflow for building and running agents.
An investigation into memory-performance tradeoffs in extreme multi-label classification (XMC), where label spaces reach hundreds of thousands of categories. We compare global and per-label weight matrix pruning strategies across four benchmark datasets, finding that global thresholding consistently preserves more predictive signal at equivalent sparsity levels.
Case study for a smart doorbell and camera security system focused on facial recognition, motion awareness, and mobile notifications.