A lecture is not a collection of knowledge
A note on why preparing a lecture starts with understanding the audience's real problem.
Long-form articles about AI and developer infrastructure, alongside shorter diary entries about building, learning, and changing my mind.
A note on why preparing a lecture starts with understanding the audience's real problem.
A note about open source, book reviews, MVP work, and why taking time off has been harder than expected.
How Co-op Translator v0.20.0 moved from simple progress output to a Rich-based CLI experience with headers, tables, progress, and plain fallbacks.
What Co-op Translator v0.20.0 taught me about separating human-facing CLI output from machine-readable structured events.
Turning a CLI-first translation tool into an agent-friendly MCP workflow changed how I think about API boundaries.
I started a 2026 series revisiting document-grounded AI systems, beginning with RAG architecture, Azure vs open-source stacks, and when fine-tuning makes sense.
Hardening Co-op Translator after real Markdown failures in multilingual documentation.
Why translation maintenance started to look more like dependency management than content generation.
Explaining Co-op Translator v0.8 as a shift from manual translation to continuous repository maintenance.
Presenting Azure AI planning decisions before writing the first line of implementation code.
Turning Azure OpenAI teaching material into a beginner-friendly Microsoft Korea video.
A compact starter guide built to help new developers reach their first working Azure OpenAI loop.
How a translation automation tool grew into Microsoft open-source infrastructure.
Presenting Co-op Translator as a full workflow: setup, CLI usage, costs, automation, and maintainability.
Framing multilingual documentation as an engineering workflow rather than a translation convenience.
Presenting Co-op Translator through the Phi-3 Cookbook accessibility case study.
The Phi-3 Cookbook case study moved my public work from Azure AI tutorials into documentation infrastructure.
Fine-tuning is not complete until the model has been evaluated for safety and quality.
The low-code companion path for fine-tuning and integrating custom Phi models.
A code-first guide connecting dataset preparation, Azure ML, deployment, and Prompt Flow.
The Azure AI tutorial that shaped my public writing around document-grounded AI systems.