Our notes.
From the people building Danubio's client work, on engineering, the industry, and what our experience teaches us.
How we rebuilt a backend without migrating the data
The riskiest rebuilds change code and data at once. We kept one shared data layer, got read parity for free, and deferred schema changes until after cutover.
More posts
Do you actually need Kafka?
More than 80% of the Fortune 100 run Kafka, so teams reach for it by default. We run it at 20,000 events a second, and we also turn it down. How we decide.
How we shipped HomeSearch AI in five months
Inside Real Estate committed publicly to launching AI-powered search in five months. The models existed; the platform did not. We built it and hit the date.
When your AI-built prototype has to scale
AI writes most of the code in a quarter of new startups. That proves the product; the architecture still has to be built. Rebuilding an AI-built MVP for scale.
What eight years on one codebase taught us about M&A
We built dashCMA, watched Inside Real Estate acquire it, and stayed on as the engineering team. Eight years on, the same team runs it for 350,000+ agents.
How we rebuilt a live multi-tenant SaaS without downtime
We took ownership of a live real estate platform and rebuilt the backend under it, moving tenants one at a time with no downtime. It now runs 5,000+ tenants.
Many of these posts started as client projects.
Tell us what you're building.