Atruvo · Senna Semakula
I diagnose and fix cloud platforms and AI systems that fail in production.
I work with engineering leaders when outages, rising costs or unreliable AI are getting in the way. I find the cause, fix what matters first, and leave your team able to run it.
Selected results
What changed after I worked on it.
20k → 80k
requests per second
Four times the throughput on the same platform.
I worked on the redesign of an event-driven platform, adding autoscaling and improving observability so it could carry four times its previous throughput. Verified under load testing and sustained in production.
Read the platform case study£50K
a month off observability costs
Measured across a year, not one billing cycle.
Reduced unnecessary metric cardinality, adjusted retention and removed over-provisioned resources on an observability platform.
Read the cost case studyResumes
after a failure mid-run
A ticket classifier that survives its own failures.
Built an automated classification workflow with concurrency controls, crash recovery and versioned prompts, so a mid-run failure resumes instead of corrupting the batch.
Read the classification case studyVideos
Production problems, explained.
One failure at a time, and what I would check first. About a minute each.
Notes
How I investigate a production failure
The measurements and decisions behind two fixes.
Six services down. 21,000 cache errors. The cache was idle the whole time.
The errors named the cache. Its metrics pointed elsewhere. Here is how I separated the symptom from the cause.
Read the noteAI code reviewCopilot reviewed my PR and invented a bug
A convincing review started with a false fact. I checked it, then found the same kind of mistake in my own investigation.
Read the note
Senna Semakula
Founder, Atruvo
Working with me
What you get
- A clear explanation of what is failing and the evidence behind it.
- A sensible order for the fixes, with the trade-offs explained.
- Changes checked against real user outcomes, not just green dashboards.
- A view of where platform spending is useful and where it is waste.
- A handover your team can use without relying on me for every change.
How I work
Diagnose
I look at the live system and test the likely causes. We agree what matters first and what evidence will show that it has improved.
Fix
I make the agreed changes with your team, check the result and document how to recover if a change goes wrong.
Handover
Your team gets the reasoning, the changes and the operating notes needed to run the system.
You work directly with me.