Stanislav – Java, Spring Boot, Grafana

A HIPAA-regulated healthcare platform built on a microservices architecture, consisting of 30 services and supporting up to 15,000 requests per second at peak. The platform was designed to provide scalable, secure, and reliable healthcare services while handling high-volume traffic and sensitive data.

• Led a 10-person cross-functional team on architecture decisions, design docs, sprint estimation, planning, code review, and on-call for a 24/7 production platform;
• Designed and shipped new services from scratch, including an AI document and speech recognition service and Keycloak-based authentication;
• Built the Kafka event exchange for the 30-microservice platform, including topics, partition keys, event schemas, idempotent consumers, retries, and dead-letter handling;
• Mentored 3 engineers from middle to senior and owned technical discussions with the US client;
• Led the migration of 30+ services from Java 8 to Java 21, reducing memory footprint and increasing concurrent request capacity;
• Instrumented services with Micrometer, Prometheus, Grafana, and OpenTelemetry tracing to reduce mean time to resolution;
• Implemented Resilience4j circuit breakers, timeouts, and retry budgets around unstable downstream dependencies;
• Designed and built AI image-to-text and speech-to-text features using Amazon Textract and Amazon Transcribe;
• Wired SAST/DAST scanning into the pipeline and remediated critical CVEs;
• Ran Gatling performance testing and right-sized capacity based on results;
• Improved CI feedback time with parallel Jenkins stages, build caching, and Testcontainers;
• Built internal MCP servers and Claude Code skills for test scaffolding, release notes, and code-review checklists.

Data platform and analytics product with a streaming event pipeline on Google Cloud.

• Led a team of 3 backend developers and coordinated delivery against the architect and stakeholder commitments;
• Redesigned the database schema and moved processing to an asynchronous, event-driven model, reducing server response time.
• Scaled the streaming pipeline to 5M+ events per day on autoscaling Apache Beam jobs in GCP Dataflow;
• Designed the Spanner data model around interleaved tables and a secondary-index strategy to keep read latency low;
• Adopted Confluent Schema Registry with Apache Avro to reduce Kafka storage and improve network throughput;
• Re-architected bulk processing into streaming, chunked batch jobs and reduced memory consumption.

An AdTech and marketing platform with seasonal peak traffic, such as Black Friday.

• Handled 5x Black Friday traffic with no incidents by using Gatling load tests, Redis caching, and horizontal pod autoscaling;
• Reworked Elasticsearch query structure and indexing for read-heavy workloads, improving query speed and reducing cluster load;
• Tuned G1 garbage-collector behavior under load to reduce GC pauses and CPU utilization;
• Moved 10 GB of cached JSON data from disk to Redis and improved system response time;
• Kept end-to-end event latency under 200 ms at peak by tuning consumer batch sizes, rebalancing partitions, adding backpressure, and stabilizing Cassandra;
• Participated in on-call rotation and incident response for a 24/7 platform with SLO/SLA targets.

Crypto-trading platform connected to multiple exchanges.

• Built fully non-blocking I/O pipelines with Netty and RxJava, keeping decision latency under 50 ms across 30+ concurrent crypto-exchange connections;
• Improved SQL-layer performance through Hibernate second-level caching.