I’m Boris Gorelik. I’ve been building data-based models since 2001, and I’ve built and led the teams that build them.
Right now I’m taking consulting and advisory work. Write to me: boris@gorelik.net. My formal CV is here, and I’m on LinkedIn.
What I do
Machine learning for problems where the answer has to survive production. Product data science first, mostly at Automattic (WordPress.com and Jetpack, 2014 to 2020) and as a freelancer since:
- Content recommendation systems for WordPress.com, for Tumblr (as a freelancer), and for a gaming company I can’t name.
- Experimentation. I led a change in how A/B tests were designed and analyzed at a platform serving millions of users, and product teams across the company adopted it. I consult other companies on the same thing.
- Churn prediction models at WordPress.com that fed product decisions.
- Social network analysis: interaction graphs built from raw activity data, for community recommendations and people analytics, covering company employees and community members across several platforms.
- Time series forecasting for financial back-office data and for an ag-tech company.
Then the systems that had to make money or catch someone:
- A dynamic pricing engine for a major Israeli supermarket chain. It was deployed and making money before the grant that was supposed to fund it was even decided.
- Multi-million-shekel Israel Innovation Authority grants for clients in retail tech and ag-tech. I designed the models, wrote the proposals, and defended them in front of the examiners.
- The real-time anomaly detection system for Automattic’s financial and operational metrics.
- A semi-supervised fraud detection model at RSA that turned into a US patent application.
I was Head of Data at MyBiotics, where I built the data team from zero. Earlier, seven years of bioinformatics and clinical trial simulation in biotech, and a PhD in computational chemistry from the Hebrew University.
I’ve hired and mentored PhD-level data scientists, and I’ve been fractional CTO for three pre-seed startups. But I don’t manage teams for a living. I build things and I tell people what will break.
I write about it
Most of this site is about data, decisions, and how to talk about them. Start here:
- Don’t study data science as a career move; you’ll waste your time! (2017)
- When your LLM pipeline silently returns zero
- Anthropic: great research, not-so-great graph
- The worst way to visualize geographic data
- The tech changed, the questions barely did.
More under career advice and data visualization. Everything else is on the blog.
I teach and speak
I teach M.Sc. courses at the Azrieli College of Engineering in Jerusalem, and I run workshops on data visualization, data-based decisions, and AI-assisted programming, in English, Hebrew, and Russian. See the teaching page and the lecture and workshop proposals.
Colleagues, friends, and the occasional stranger ask me for advice. I’m always happy to help. For anything longer than a coffee there’s mentoring, and for founders there’s strategic advisory.
Other things I do
I produce This Week in the Middle East, the oldest active Hebrew-language podcast devoted to the region. These days I’m the sole producer.
Talks, lectures, and workshop recordings go on my YouTube channel.
On the side, I’m building Loud Camel, a tool that helps researchers get their work noticed by people and by AI search engines.
My academic publications, ORCID, and Google Scholar profile are on my research page.