I am a software engineer focused on backend distributed systems, AWS-based services, observability, resilience, and practical AI-agent tooling. This site collects projects, demos, and sanitized engineering case studies.
Projects I can demo publicly or safely explain with sanitized data. The theme is practical software: small enough to ship, useful enough to keep using.
A lightweight personal tool for badminton drop-in booking, built around a real recurring need instead of a toy problem.
A local knowledge workflow that lets an AI assistant retrieve durable context from a structured note vault without stuffing everything into the prompt.
A workflow prototype for classifying recruiter messages, job postings, and follow-ups into structured job-search actions using n8n-style automation.
A proposed guardrail for AI agents that intercepts risky side-effect tools, such as sending messages or writing outside approved paths, before execution.
A verification loop where tasks define pass/fail criteria, the agent checks its own output, and failures trigger targeted correction instead of vague retries.
A sanitized architecture sketch for the personal knowledge and assistant workflow behind my AI tooling experiments.
The key idea: keep private knowledge in a local vault, index it, expose controlled retrieval tools, and let the agent cite or summarize relevant context only when needed.
Production engineering stories written without proprietary details. These focus on the engineering patterns: safety, observability, rollout, and reliability.
How to approach an index migration with mapping validation, dual-write checks, traffic replay, monitoring, and rollback-oriented launch planning.
Adding service metrics, alarms, dashboards, chaos/stress testing, and killswitch visibility to improve operational readiness.
Cleaning up unused credentials, fixing rotation issues, and reducing operational security risk across multiple environments.
Increasing regression confidence by expanding integration coverage across more processors and workflows before deployment.
I am interested in backend infrastructure, AI-agent platforms, tool safety, evaluation-driven automation, and pragmatic systems that make daily work easier. Outside work, I play badminton and guitar.