Journal
Writing that shows how I think through product direction, UX decisions, architecture tradeoffs, and practical AI workflows.
Journal snapshot
Current writing archive
0
Entries
0
Topics
0
Current view
Sep 29, 2026
Latest
All entries
Filter by topic to scan writing on AI workflows, product decisions, system design, and lessons from building.
> After building an agent, how do we know it works well, rather than just feeling good? My AI assistant is meant to respond as me. Previously, when I asked it “Who are you?”, it would answer in that role, introducing itself as me. After DeepSeek updated its LLM, however, the assistant began answerin
> Managed Agents are easier to understand by starting with a local Agent. Once it needs to serve more users, run for longer, and recover reliably from component failures, its execution loop, durable record, and execution environment need to be separated so that each can be recovered and replaced ind
> An agent harness is the software around an LLM that runs the execution loop, executes tool calls, and manages the context and state needed to continue a task. The model decides what to do next; the harness coordinates how that decision is carried out. To understand why these pieces are needed, I f
A tool is an external capability an AI agent can call to perform actions or access information beyond what the model can do by itself. Designing an effective tool means making it clear when to use it, how to call it, and what the agent can learn from its response. LLM-oriented design In traditional
I used to think that improving an AI application mainly meant writing a better prompt. That was mostly true when the task involved one request, one model call, and one response. Once an agent begins working across files, tools, retrieved documents, and a long message history, however, the prompt bec
The word is often used for almost any application that combines an LLM with tools. I used to think that the more steps and tools a system had, the more "agentic" it was. What matters more, however, is who decides what happens next: predefined code or the model itself. I wrote this note to clarify t
Kubernetes is an open-source system for running and managing containerized applications across physical machines, virtual machines, and cloud instances. The reason it exists is simple: once an application is split into many microservices, containers become easy to create but hard to operate. Kuberne
A RAG system is easy to demo but hard to trust. The first version of my personal AI Q&A system worked well enough to feel useful. It could retrieve context from my portfolio, project notes, technical writing, and GitHub-related sources, then answer questions through a conversational interface. But o
Background GitHub Actions provides a generous free tier, but it comes with specific limitations. The free usage is typically 2,000 minutes per month for standard GitHub-hosted runners. If your team has a bunch of repositories or jobs to run, you may quickly meet this limit. So what should we do afte
I. The Problem with Static API Documentation Your team has done the hard work. The API endpoints for the next big feature have been designed and agreed upon. But where does this crucial design live? All too often, it's trapped in a Google Doc or a PDF. While this is better than no documentation, it
> GraphQL is a query language for APIs, and a server-side for executing queries using a type system defined for data. Introduction of RESTful APIs In a typical software project, a RESTful API (Representational State Transfer Application Programming Interface) serves as a standardized way for diffe
> A Cloud Development Environment(CDE) is a type of software development workspace that runs in the cloud, rather than on your local computer. It allows you to write, run, and debug code entirely through a web browser from any device. Background Recently, my MacBook Pro has started having issues th