API response times halved.
Instrumented diagnosis and targeted changes, not a rewrite.
SRE · Backend · Production systems
Make production reliable. Cut costs. Unblock delivery.
I work across backend and infrastructure on high-traffic production systems to diagnose problems, deliver the fixes and leave the team able to operate independently.
Eight years building and running complex systems, with end-to-end ownership from diagnosis to operations.
99.99 %Uptime over a full yearEnd-to-end SRE ownership · 30M requests/day
10× lowerp99 latencyThe average showed no issue
3× lowerInfrastructure billCompute moved to bare metal · egress removed
When to call
When to call Solidnines
When a production issue needs ownership, not another recommendation deck.
Production is fragile or unpredictable
Incidents recur, recovery depends on a few people, or reliability is not measured clearly.
Latency or failures are rising
The averages look acceptable, but users, tail latency or background jobs say otherwise.
Cloud or CI spend is drifting
The bill grows faster than usage and the trade-offs are hard to quantify.
A migration has become risky
Kubernetes, PostgreSQL or cloud changes are blocked by production risk.
The problem crosses several layers
Backend, PostgreSQL and infrastructure all look like plausible causes. I measure the system end to end and fix the problem where it actually is.
Reliability lacks a clear framework
Observability, SLOs, incident response and production responsibilities remain implicit or scattered.
How I work
Measure. Fix. Hand over.
The right solution, not the most impressive one.
Measure
Establish the signal, the baseline and the constraint before choosing a solution.
Fix
Make the necessary changes, one controlled step at a time, and verify the result.
Hand over
Document, automate and transfer the operating knowledge. The team carries on without me.
Results
Other measured outcomes
Failed background jobs reduced by 95%.
CI brought down from over ten minutes to under three.
The CI bill was also cut to a third.
PostgreSQL version upgrades and sensitive maintenance delivered without downtime.
Including pg_repack and Row Level Security changes.
Expertise
Three areas, one production system
Work is organised around the problem to solve, not around a tool catalogue.
Production reliability and performance
Observability, SLOs, incident response, profiling and end-to-end latency analysis.
Backend and PostgreSQL
APIs and services in Python or Go, asynchronous processing, PostgreSQL, performance and sensitive production migrations.
Infrastructure and delivery
Kubernetes, cloud, bare metal, CI/CD, infrastructure as code and cost control.
Frequent environments — Python, Go, PostgreSQL, Kubernetes, AWS, GCP, OVH bare metal, Terraform/Pulumi, Cloudflare, Datadog.
A simple first step
Start with a production diagnostic
A short, focused engagement to establish what is happening in production and determine where to intervene before making broader changes.
Objective diagnosis
Signals and constraints are measured, not assumed.
Causes and risks
The main failure modes and technical risks are made explicit.
Prioritised actions
Changes are ordered by impact, risk and effort.
Clear next step
What should happen next, including when my involvement is not needed.
No commitment to a longer engagement.
Start with a production diagnosticA production problem to solve?
Describe the problem in a few lines. I reply within two business days and will tell you plainly whether I am the right person.
Talk it through
Thirty minutes, no commitment. Most useful if you already have a number or symptom to examine.
Talk about your production — 30 min