Yan Myo Aung — AI Software Engineer | Full-Stack & Intelligent Systems. I’m an AI Software Engineer and full-stack developer focused on building intelligent software systems. My work combines AI agents, LLM applications, backend engineering, and system architecture — from multi-agent workflows and AI tool integrations to RAG pipelines, model fine-tuning, scalable APIs, and production-ready applications.

AI agentic systems, cybersecurity platforms, developer tools, and full-stack enterprise systems — 15 projects including InnoIgniterAI, Trace, K.I.N.E.T.I.C., and Stealth Copilot.

What are his core technical areas?

Agentic AI, LLM systems, AI infrastructure, cybersecurity, and full-stack system architecture, with experience across Python, TypeScript, Rust, Go, and modern AI tooling.

What makes his projects different?

He focuses on complete, production-oriented systems rather than isolated AI demos, combining AI with architecture, security, observability, testing, performance, and deployment.

Is he available for remote work?

Yes. He is open to international remote opportunities, particularly AI Software Engineering, backend, full-stack, and AI infrastructure roles.

How can I contact him?

Email ymyo44277@gmail.com — GitHub and LinkedIn are available through the profile links. Based in Myanmar (UTC+6:30).

Writing

Notes & case studies

Short notes on what shipped, what failed, and how it was measured — AI agents, security, and systems.

7 posts · newest first

2026-10-1410 min

Guardrails Belong on the Critical Path

How I built a 3-tier safety interceptor for an AI agent framework — one chokepoint between planning and execution, errors-as-data, default-deny approvals, and the honest limits of regex blocklists and LLM judges, audited line-by-line against the 0.3.4 code.

AI AgentsSafety & GuardrailsModusAIAgent Infrastructure+1

Read → /guardrails-belong-on-the-critical-path

2026-10-106 min

Burst-Proofing a SQLite Writer: Measuring Spill-to-Disk Overflow

What a 20,000-event burst taught me: one SQLite writer commits 7-9k ev/s while overflow producers crawl at 88 ev/s — with the data-loss paths left open.

AI AgentsObservabilitySQLitePerformance

Read → /burst-proofing-a-sqlite-writer-measuring-spill-to-disk-overflow

2026-09-286 min

TSE Queries Should Degrade Instead of Dying

How TSE's router merges SQLite hot and Parquet cold tiers with overlap windows, staying available when either tier fails.

GoSQLiteParquetQuery Engine+2

Read → /queries-should-degrade-instead-of-dying

2026-09-2510 min

Observability for Agents Should Cost Nothing

How I built local, asynchronous observability for an AI agent framework with a bounded queue, background writer, and SQLite—keeping tracing off the critical path.

AI AgentsPythonObservabilityAgent Infrastructure+2

Read → /observability-for-agents-should-cost-nothing

2026-08-175 min

Build Your First AI Agent in Ten Minutes

Install Modus, ask a model a question, give your agent memory, hand it a tool, and trace everything. A beginner path with runnable code.

AI AgentsPythonModusAITutorial+1

Read → /build-your-first-ai-agent-in-ten-minutes

2026-08-1110 min

Flush Should Never Make Ingest Wait

How TSE's drain-loop flusher, contiguous watermark, and idempotent writes survive SIGKILL mid-flush without blocking ingest.

GoDurabilityCrash RecoveryIdempotency+2

Read → /flush-should-never-make-ingest-wait

Showing 6 of 7 posts