Notes on data, AI, IT
and security
No marketing fog. The way I think about real problems with founders and managers.
Multitenancy and trust boundaries in SaaS
Where the convenience of a shared platform ends and the noisy-neighbour risk begins.
The platform team as the next step after strong operations
Why a centralised platform is becoming a separate function, and what that means for an IT leader.
Cloud exit plan: contract, migration, and the exit test
While everything is working, nobody calculates the cost of leaving. That is exactly why it needs to be calculated before it becomes urgent.
Model quality vs data freshness: what matters more in applied analytics
In most real tasks, the winner is not a smarter model but a better-managed operational loop that keeps data current.
Is MapReduce enough: where the batch model starts hurting business
Some business scenarios already demand a different computation tempo. Not because batch is bad, but because the latency has become too expensive.
What 2014 will demand from architecture: fewer monoliths, more discipline
The preparation is not about buzzwords - it is about faster release cycles and service connectivity. That requires specific architectural decisions made now.
Atlas and DARPA: what advanced robotics tells us about everyday automation
Robots far removed from your business are still useful as an indicator - they show how mature sensors, balance algorithms, and human-machine interaction have become.
Personal data compliance is not just a legal problem
Data architecture and access controls directly shape legal risk. Technical decisions made today will become legal problems tomorrow.
FSTEC Order 21: what it changes in the practical protection of personal data
The Russian market gets a more concrete language for protection measures, security levels, and threat classes. What this means for companies that process personal data.
Spark, Hadoop, or MPP: choose the workload type, not the brand
Different tasks need different computation and storage modes. Getting the platform wrong costs more than it appears at the start.
Technical debt in data pipelines: why \"we will rewrite it later\" almost never happens
Data pipelines age faster than they appear to, especially when nobody owns the schema. Deferring the refactor has a concrete cost.
Incident communication: who says what while the technical team still has no answers
Reputational damage from an incident begins before the technical answer arrives. The communication plan needs to exist before it is needed.