What's up?

TJ Chalokia here.

This is a personal site about agents, memory, research, taste, and the operational side of becoming harder to fool.

Some posts are practical: context engineering, evals, deep research, system design, and how agents should remember things. Others are personal: ambition, discipline, taste, health, attention, and the quiet machinery of rebuilding yourself. That split is intentional. The technical work and the life work are not separate tabs in the brain.

Start Here

• The Context Window Is Not Your Memory Architecture
• Giving Agents a CLI Is Not About Commands
• User Manual
• Resume

What This Site Is

This site is where I put things worth keeping:

• practical notes on building AI systems that survive contact with reality
• essays about memory, research, agents, and operating systems for thought
• personal writing about ambition, self-respect, pacing, and taste
• field notes from whatever I am building next

The goal is not to look busy.

The goal is to become legible.

What I Am Building

The working idea behind this site is a life engine: a broader operating system for notes, essays, experiments, and hard-won taste from building with AI. That phrase is useful as an internal concept, but the public site should first read as a person, not a project brand.

The writing should feel like a workshop with clean benches. Open enough to learn from. Organized enough to trust.

Current Thesis

Most people treat AI as a tool problem.

I think the deeper problem is state.

What survives the prompt? What stays true after compaction? What can be audited? What does a model know because it saw evidence, and what does it merely repeat because the transcript got too persuasive?

That is the territory I want this site to claim.