Fix slow and unreliable software
Software performance and monitoring
Find slow requests and recurring errors, with monitoring that helps your team see what is happening.
Discuss this serviceI start by finding where the problem occurs. That means looking at real requests, database work and error reports before making changes.
At Bykea, removing slow operations helped bring a critical API’s p95 response time from about one second to 200 to 300 milliseconds. At PractiCal, I optimized MySQL queries, integrated Sentry and supported production issues on call while leading the technical team.
Observability helps your team follow what happens inside an application. I can help set up logs, performance traces, dashboards and alerts using Elastic APM and the ELK stack, Grafana or Sentry. We choose tools that fit the system and use the measurements to review the changes.
Tools and technologies
We choose the tools that fit your project and existing systems.
Elastic APM
ELK stack
Grafana
Sentry
What the work can include
- Investigation of slow requests and recurring errors
- Database query improvements
- Observability and application performance monitoring (APM) with Elastic APM, ELK, Grafana, and Sentry
- Dashboards and alerts for application health
- Measured changes and production support
How the work runs
- Look at the symptoms, affected users, and recent changes.
- Review monitoring, logs, and slow operations to find the cause.
- Agree the fixes and deliver them through the normal review and release process.
- Compare the results with the original measurements and document the changes.
Questions about this service
What if we do not have monitoring yet?
Then we first agree what needs to be measured. I can help set up tools such as Elastic APM, Grafana, or Sentry so the investigation is based on evidence.
Will you need production access?
We agree the access required for the investigation. Monitoring data, logs, and a suitable test environment may be enough; any production access or changes are handled through your agreed process.
Can you guarantee a particular speed improvement?
The result depends on the cause. I measure the affected operations and explain what the evidence supports before proposing changes.
Is there a relevant example of your work?
At Bykea, removing slow operations brought p95 response time for a critical service from about one second to 200 to 300 milliseconds. The Bykea case study explains my contribution.
How engagements work
What to expect
when we work together.
Payment terms are agreed before the work begins.
Ways to work
A fixed project, ongoing development, or extra capacity for your existing team.
Replies
I reply to messages within 2 to 3 hours.
UK hours
I work on GMT+5, and three hours of my day overlap with the UK working day.
Code and payment
The code stays mine until the work is complete. You receive it once the final agreed payment is made.
Related work

Payments and delivery engineering at startup scale.
Led engineering for Payments and Delivery, from pricing and location features to production performance.
Senior Software Engineer at Bykea
Leading the refactor of a live meal subscription platform.
Led a Laravel refactor and production support. The refactor reduced production bugs by approximately 90%.
Delivered through Technyx Systems