Silent errors
Someone who doesn’t know what a value means won’t notice on the day it comes through wrong.
01Will it hold when the volume gets serious?
Moving data is easy. Keeping it reliable at scale is harder. We build pipelines where cost, runtime and failures stay visible, even when the volume grows.
Scale isn’t the hard part
Data volumes are rarely the problem. The problem is what happens when the volume rises: cost, runtime, and whether anybody notices a failure.
A pipeline that works in a test is craft. One that holds under real load for years is a discipline: retries, ordering, back pressure, and a team that understands the data.
Why no client names2x10
Our clients hire us to build what gives them an edge. Confidentiality is part of the job. Publishing the details could give that edge away. We’ve delivered in regulated environments and on projects worth hundreds of millions.
What we've built
Our own company, not a client project. Novid20 was the non-profit COVID-19 start-up that later became 2x10.
We built and ran the whole chain: the test interface used by schools, nine lab connections in Bavaria alone, the processing, the notification going back out. When a lab interface failed at three in the morning, we fixed it at three in the morning.
A processing system for a client, part of their core service. New data arrives and the pipeline starts itself.
It scales up to hundreds of compute cores under load and back down to nothing, which is the half of scaling that actually saves money. Thousands of analyses run in hours rather than weeks.
Our Story: Novid20. COVID start-up that became 2x10.
0135M+COVID tests
025,381institutions
033.5M+tests per month
049+labs connected
Domain knowledge
Understanding the data, not just moving it
Confidently wrong
A pipeline that reliably passes on wrong values does more damage than having none at all.
02Domain fluency
We learn the domain until we can discuss the data with your own specialists.
03Production questions2x10
What makes them hold up
A pipeline is judged on its worst night, not its average one. Three questions cover it.
- Does it scale back down?Scaling up is easy. Scaling down is what saves the money, and most systems never learn it.
- What happens to item 4,117 when 4,116 fails?Every item carries a state you can inspect, and failures wait for a second attempt instead of disappearing.
- Who notices when nothing arrives?We monitor silence as well as errors: zero events looks exactly like a quiet Sunday, until someone realizes nothing has come in since Friday.
What we could build for you
- We process large volumes without your costs growing at the same rate.
- We connect the systems and interfaces that are bridged by hand or by spreadsheet today.
- We rebuild existing processing so that peak load stops being an event.
- We make it visible when something has not arrived, before your customer is the one who tells you.
- And whatever you have in mind that AI and solid software engineering could make possible.
Under the hood
Architecture · event-driven processing · queues and ledgers · idempotency · recoverable state · Scale · auto-scaling · spot compute · columnar storage for fast queries · Operations · infrastructure as code · observability · silence detection
Most engagements start small: talking it through, a few weeks of engineering, something real you can judge. If it works, we take it further.If it doesn’t, you find out early
More work2x10
AI Content
Quality at volume
View themePrivate AI
Your data stays
View themePlatforms
Built to last
View themeN° 06 Contact
Tell me what you’re planning
No form, no ticket system. Email me directly and you get me, not an inbox someone else manages.
Moritz Miedler · Founder, replies personally2x10 Technologies · Viennahello@2x10.com