Konstantin Tchernov

I started at Whip Around as one of its first developers, and the SaaS company saw fast growth. I rearchitected the monolith back end to a scalable multiservice system to cope with growing platform volume and changing use cases from new, more demanding customers.

When we still saw a small number of unacceptable failures reported through our customer success team, I conceived and won approval for a project to improve the reliability of key parts of our user flow.

Led the project to add new metrics and more fine-grained SLOs to get us the extra reliability we needed in the company's core area of business. I enhanced our suite of cloud metrics, introducing detailed performance tracking, front-end tracking, and security indicators.

I took that further than cloud system metrics—to tracking failures that a user might see through business intelligence data. We could then detect failures on a user level that may not have been obvious just from cloud performance metrics. This was a coordinated effort to combine data from various front-end channels (web and mobile) and back end.

A greenfield system to dynamically generate templated organizations inside Xero, a global accounting SaaS platform. This would allow university lecturers to create templates for accounting organizations and invite 300+ students at a time to the platform, each student getting their own sandbox.

This required connecting to many different subsystems within Xero, and it had specific performance requirements to allow spinning up these large datasets on-demand and quickly.

I rearchitected the platform to perform these heavy organizations on back-end workers that would scale with the load while rate-limiting to not place too much load on the rest of the internal services at Xero. As it was an accounting platform, there were also many strict privacy and security requirements, SOC 2 audit compliance, and GDPR compliance.

My new architecture put the generation of these organizations on a background worker queue. I also worked with several teams within Xero across multiple countries to give us the right tools to access their secure services.

This resulted in a successful product with a much higher performance than we could proudly demo live with large datasets.

A private investment research platform with stock tracking and private research articles. This was subcontracted from the client to my agency.

I made architectural adjustments to the platform for improved performance and reduced cost. I also gave the developers much better tooling to maintain the platform.

The cost was a big concern for the client, so I implemented a proposal to move to managed containers (AWS Fargate), removing the need for software updates to the underlying hosts while setting up timed regular deployments of containers to pick up security patches frequently. I also introduced and tuned auto-scaling to keep costs minimal while supporting the client's growth.

With several other proposals I made, the result was a 25% decrease in cloud hosting fees and reduced person-hour fees in platform maintenance.

I moved the core services to use trunk-based CI/CD in GitHub Actions on the developer side. The Docker containers that were built could then be pulled and run by developers for local development, reducing the disparities between local and cloud environments. I also won client approval to pay for Datadog to enhance the monitoring for this project.