Roni Kobrosly Ph.D.'s Website

I was invited to speak at the Prefect-sponsored PyAI DC event and had a wonderful time meeting DC's AI community. It was an honor to speak alongside both AWS and government data leaders. The highlight of the night was Prefect's leadership giving some details about the upcoming major update of FastMCP...

Okay, I decided to take that free-floating ASCII diagram code and make it into a proper open-source python package. I present to you “RetroFlow”. On the surface it seems silly, but there are a lot of advantages to using ASCII flowcharts, particularly in this age of agentic AI...

I forget how it started but recently I've been enjoying making these cool retro-ish, ASCII flow diagrams for my personal projects. They look like something you'd find in a dusty cabinet of early NASA documents. Code linked in post...

It's been about a year since I launched Data Compass AI, and it's collected over 3,000 survey responses from organizations spanning finance, retail, and more. For various reasons I'm planning on winding down this experiment. I wanted to give it a proper goodbye and take a moment to share what the data tells us. It also gave me a chance to try out some agent-generated (and human-verified) analytics.

Over the last few years I've been working within a large SRE org and I've really come to deeply appreciate observability in the software engineering world. I'm a data science / machine learning / AI engineer, and sadly AI/ML observability feels like a niche topic. I figured it was time to put proverbial pen to paper, and write up some of what I've learned. This is the first installment of a three-part post on observability, OpenTelemetry (Otel), and ML microservices. Part 1 focuses on an examples of Otel to a classic ML microservice, Part 2 is using Otel trace data to perform automated root cause analysis, and Part 3 will focus on observability for large language models (LLMs)...

The wonderful terminal integration and easy MCP configuring of Claude Code has gotten me really excited. I would even say it fully restored my joy for development that I had a decade ago! I've been playing with it for a couple of months and think I finally have a claude.md file that maintains the quality of the AI's work between sessions.

Data Compass AI is now fully in production! 🎉. After weeks of work in the little cracks of free time I have these todays, I'm launching a new web service to help our struggling data science / analytics / AI data community.

I'm so close to delivering a production web service to help folks in the data community. It's the most ambitious tech thing I've done and I'm extremely excited and nervous! I learned a lot along the way...

Over the holiday break, I had the opportunity to try out the Cursor Pro, one of the newer and more discussed GenAI coding assistants and IDE. It is meant to be like GitHub CoPilot, but the idea is it is its own IDE and AI is infused into each aspect of it. I learned a lot about both Cursor and the python Django framework through Cursor.

I recently posted on LinkedIn about the state of "data maturity" at organizations and got a lot of feedback. Are organizations, generally, more data mature now than say a decade ago?