Our Story

After years building ML systems, I kept seeing the same pattern across every project: feature engineering consumed 80% of the time and resources.

The personalization paradox: Start small and you're stuck with black-box solutions that don't scale. Go big and you're looking at 6+ month infrastructure projects with massive teams.

Every team I advised faced the same choice:

• Option A: Use managed rec-sys solutions — great for MVPs, but they hit hard limits at scale. Unpredictable costs, no control, eventual migration pain.
• Option B: Build custom systems — better performance, but requires data pipelines, feature engineering, ML teams, and months of R&D.

Both paths forced a painful migration when growth happened.

The "aha moment" came during a project audit. We discovered that 80% of our effort went into feature engineering — building data pipelines to support it, cleaning data, creating embeddings, handling temporal patterns, sorting user histories. Only 20% went to the actual ML models that generated recommendations.

What if we could eliminate that entire layer?

That's why I built dodo — economic infrastructure for personalization that removes the feature engineering layer entirely. A unified platform that starts as zero-ops serverless recommendations and scales seamlessly to enterprise-grade, low-latency infrastructure.

The "serverless-to-scaled" spectrum isn't a feature — it's our philosophy. Developers shouldn't have to choose between "easy to start" and "powerful at scale":

• Start: Upload raw user data — shopping histories, content preferences, behavioral logs. No preprocessing required.
• Scale: As traffic grows, get dedicated resources, custom models, or sub-50ms latency — without changing a single line of code

This creates a growth loop where developers focus on building their business, not engineering features.

Why now? Personalization has become as fundamental as payments. Every app needs to recommend relevant items, but teams waste months on data preparation instead of building products.

dodo bridges that gap. We handle the feature engineering and ML complexity so developers can focus on what matters: growing their business.

The vision: Make state-of-the-art personalization as accessible as modern payment APIs. Start with raw data today, scale to millions of users tomorrow — one platform, zero migration.

Want to see it in action? Join our beta and make your first recommendation in under 5 minutes.