Agents, and what I've done with them

It has become apparent that coding agents are able to do things good enough that we must use them, and move up a layer of abstraction. Tools like Agentation, Pi or openclaw are already proving to be the next generation of tools that the "programmer" must use.

Tools seem to be the next step in the LLM's journey to become more intelligent. Every noticeable achievement of the modern LLMs has come from them benefitting and leveraging powerful tools like bash/grep etc ... (for coding agents) or Lean4 (for mathematical proofs). Some might even wonder that harness refinement may just take us to the next step of intelligence explosion.

Anyways, I wanted to list some projects that I've undertaken, which wouldn't be possible without our modern tools that are LLMs:

• Created a small financial DSL, and a pipeline to create, backtest and deploy a financial algorithmic strategy

• Start writing lean4 kernel for a deep learning theory I found intriguing, and wanted to formalize

And that's it ! I myself expected this list to be longer, but these are the coolest ones I can imagine. I'll keep expanding this list for sure !