About

I'm a data scientist in Tel Aviv working on credit risk and fraud. I came to machine learning from accounting, and I still think of a model as an entry in someone's books.

From the ledger to the model

I started at UHY Israel as a financial auditor, auditing public and private companies, and spent time as the controller of a listed financial-services firm. That work teaches you what a number goes through before it reaches a report, and what it costs when it's wrong.

In 2018 I moved to Dov & Co, a CPA firm, as a software developer and blockchain administrator. I built financial software and blockchain applications, and advised clients on how digital assets are regulated and taxed. Writing code to automate the work I knew from audits led me to data science.

In 2020 I completed the Israel Tech Challenge data-science fellowship and joined Fido Credit Solutions, where I've built credit-risk models for three international markets, fraud rules, uplift models for Collections, forecasts for Finance and, lately, agentic AI workflows that take the repetitive parts out of model development.

What I bring to a team

A model in lending is a business decision written in code. I own the whole path, from features and evaluation to deployment and monitoring. I also explain the result in terms that finance, risk and compliance already use: losses, margins, controls and audit trails.

Away from work

I play competitive chess in a local community league. It's good training for a risk modeler: you're always weighing the move that looks best against the one that's least likely to lose.

I also build things on the side. The Global Min is my data-science interview-prep app.

Languages

Hebrew (native) and English (native).

Get in touch

The best way to reach me is email, at yuvalherm@gmail.com. I'm also on LinkedIn.