Building ML Systems & Agentic Workflows.
Machine Learning & Edge AI engineer focusing on signal-theoretic metrics, real-world deployments and agent-in-the-loop systems.
ML & Edge AI engineer working on agentic workflows, robustness, and real-world systems.
Combining classical signal processing with deep learning to make agents and LLM systems stable enough for messy real-world inputs.
Featured Project
JumpNet — Vision-based Game Agent
A one-button game agent that learns when and how long to jump from raw gameplay frames. JumpNet uses a CNN backbone with a dual-head output (classification + regression) trained on recorded play sessions, and is deployed in a small, GUI-driven simulator.
Links
Active repositories for ML agents, edge deployments, and small tools that I use in day-to-day workflows.
Open profileCareer timeline, talks, and short posts about ML systems, edge AI and agentic workflows.
ConnectDeep dives into ML experiments, debugging stories, and end-to-end project breakdowns.
Read postsWavelet-based feature engineering, clustering for parity detection, and future work on logit stability metrics.
View papersI'm happy to talk about ML systems, edge deployments, or potential collaborations.
Send email“From Human-in-the-Loop to Agent-in-the-Loop” — a practical look at evolving ML workflows.
Go Webinar