Ran Aroussi is a software engineer and founder building agent infrastructure and developer tools. He created yfinance and quantstats and currently works on Resident AI at VarOps and CTO-as-a-Service through Automaze.[2]
Start with Git Is the Wrong Granularity for Agents →
Links
Owned professional site Owned biography and current workCTO-as-a-Service · Automaze
Resident AI development · VarOps
Career
Aroussi’s owned biography describes a career spanning software, online marketing, advertising technology and automated trading. His current projects include MUXI agent infrastructure and the Cloop engineering control plane. At Automaze he works with founders and technical teams; at VarOps his focus includes organizational knowledge and AI systems. His public writing connects those projects with practical experience operating software businesses.[2]
Contributions
Keep working history and decision history at different scales
His development-history article proposes two stores beside Git. ghf captures restorable worktree states between commits, with content-addressed objects and optional synchronization. GitStory attaches cited explanations of decisions to accepted changes. The distinction reflects different lifetimes: experimental states can expire after work lands, while the reason for a decision may matter much later. The article presents ghf as an early release and GitStory as a private preview.[6]
Adapt a local command-line program to an API-shaped client
CC Bridge runs the Claude Code CLI and reshapes its output for clients expecting Anthropic-style requests and streaming responses. Its README describes a personal, unofficial prototype for local experiments and integration testing. It requires an installed, authenticated CLI; its compatibility claims and upstream behavior remain those of an experimental wrapper, rather than supported API access.[5]
Make market-data retrieval accessible from Python
yfinance offers a Python interface to publicly available Yahoo Finance data, including historical prices and ticker information. The project is independent of Yahoo and documents limits on the data’s intended use. Its role is data access for research and analysis; obtaining a series still leaves the user responsible for checking its coverage, adjustments and fitness for a particular question.[7]
Tools, writing & talks
Writing, research & talks
Start here
Git Is the Wrong Granularity for Agentsvarops.comAn article separating restorable experimental file states from cited explanations of accepted code changes.
Read the checkpoint and decision-story examples, then compare their retention needs with your own Git workflow. Git familiarity helps; ghf is an early release and GitStory was a private preview when the article was published.
Author[6]