Custom predictive models for organizations that care about more than just the bottom line. Built by one person who actually cares about what you're trying to do.
Let's talk →Here's the short version: A healthcare SaaS company was evaluating an enterprise ML platform — quoted at $135k/year — for churn prediction. The vendor ran their proof of technology and got 57.7% recall.
My existing custom model was already at 89.7% recall. The executive team declined the vendor on performance grounds.
My custom model outperformed a $135k enterprise vendor by 32 percentage points on the metric that mattered most.
That's not a flex. It's a demonstration that custom-built models, in the right hands, outperform off-the-shelf platforms — because they're built for your specific data, your specific problem, and your specific definition of success.
I'm now taking that same approach to organizations with missions worth protecting.
Know who's about to leave before they leave. Donor attrition, member lapse, customer churn — same problem, different label. I build models that flag at-risk accounts with time to act.
Not all supporters are the same. I build segmentation models that show you exactly who to prioritize, how to talk to them, and what they're likely to respond to.
A central reporting hub that pulls in your existing tools (PowerBI, Tableau, whatever) into one place. Set it up once, it just runs. Your team finally has a single source of truth.
For the last 8 years, I've been a senior data scientist at a healthcare SaaS company. My job: build models that predict which customers are at risk of leaving, so the customer success team can intervene before it's too late.
I'm good at it. But more importantly, I care about why I'm building it. I want to work with organizations that have a mission — not just a quarterly revenue target.
That's Pydealist. One person, custom models, mission-driven clients. No sales team, no account managers, no bureaucracy. Just me and the work.
I learn about your organization, your data, and what "good" looks like for you. No pitch, just listening.
One to two weeks. I work with your existing data, build a custom model, and deliver it with explainability (SHAP values) so you understand why it makes the predictions it does.
We integrate it into your workflow. Dashboard, alerts, whatever makes sense. You start retaining more of the people who matter.
Ongoing retainer for model maintenance, new features, or the dashboard. No lock-in. If it stops working for you, we part ways.
I take on a limited number of clients at a time. If you're mission-driven and have a data problem worth solving, let's talk.
Email me →Or just reply to the postcard that brought you here.