Event-driven • AWS • Java

Building scalable, observable cloud systems

Hi, I'm Kiryl Miadzvedzeu — Senior Java/AWS Engineer. I design and deliver event-driven platforms with Step Functions, Lambda, and DynamoDB. I love migrating legacy to microservices, reducing latency & cost, and adding solid observability.

View Projects Contact

Projects

Automotive Group — Calendar Management System

2021–2022

Internal Google-style calendar for an automotive group: schedule and manage events with filters, tags, and priorities; speaker profiles with photo + talk details; multi-user views for departments.

Java 11
Spring
Hibernate
Gradle
MS SQL
Flyway
RabbitMQ
Docker
GitLab CI
Testcontainers
JUnit
Mockito
Swagger/OpenAPI

Case

SolbegSoft / Helmes — Risk & Asset Protection

2021–2022

Protection software for automotive & property insurance. Migrated parts of a legacy monolith to microservices; delivered new features, QA hardening, and stability improvements.

Java 8
Spring
EclipseLink ORM
Maven
Liquibase
SOAP
XML (SAXIF)
Tomcat
ActiveMQ
Ruleset
JUnit
Mockito

Case

Shell Energy — Smart Metering Platform

2022–2024

Real-time metering for 2,200,000+ users. XML-only legacy monolith wrapped by adapters; traffic strangled to WebFlux microservices on Kubernetes. Observability via Prometheus/Grafana with RPS and hourly/daily/weekly ingestion metrics.

Java 17
Spring WebFlux
Spring Boot
RabbitMQ
PostgreSQL
Cassandra
Docker
Kubernetes
Prometheus
Grafana
Liquibase
Testcontainers
JUnit
Mockito
Cucumber
REST
XML
Maven
Git

Case

Control-Plane: Event-driven tenant workflows

2024–2025

Internal multi-tenant control-plane for a SaaS platform: orchestrates full tenant lifecycle (onboarding, configuration, offboarding) using AWS Step Functions + Lambda. Implemented saga patterns with compensations, idempotency, fan-out/fan-in, and async SQS jobs. Built observability with CloudWatch/X-Ray and fault-tolerant retries. Achieved −25% Lambda cost and −30% p95 latency through provisioned concurrency, batching, and payload optimizations.

Java 21
Spring Boot
AWS Lambda
Step Functions
EventBridge
DynamoDB
SQS
Terraform
X-Ray

Case

Experience

Backend Software Engineer — Godel Technologies
2024 – Present
- Event-driven workflows: Step Functions, Lambda, EventBridge, SQS
- Team setup: 5 backend developers, 1 QA, 1 DevOps, 1 Tech Lead, 1 BA, 1 ADC
- −25% Lambda cost, −30% latency via optimization
- Observability with CloudWatch & X-Ray
- Process: Scrum mixed with Kanban for delivery flexibility
Backend Software Engineer — Godel Technologies
2022 – 2024
- Smart metering platform; hundreds of microservices
- Java 8 → 17, Spring 2.3 → 3.0 modernization
- Shared libraries, 98% bug-free releases with robust tests
- Agile workflow: Scrum + Kanban hybrid
Backend Software Engineer — SolbegSoft / Helmes Group
2021 – 2022
- Risk management & asset protection for automotive/property insurance
- Migrated parts of a legacy monolith to microservices; QA hardening
- Team setup: 1 Tech Lead + 5 developers + 1 ADC
- Agile workflow: Scrum + Kanban hybrid
Backend Software Engineer — Automotive Group — Calendar Management System
2020 – 2021
- Greenfield internal Google-style calendar (events, filters/tags/priorities, speaker profiles with photo & talk details)
- Sole backend engineer in a 6-person team (1 BE + 5 FE)
- Team setup: 9 backend developers, 1 ADC, 1 Tech Lead
- Built from scratch in ~10 months + 2 months production support
- Process: Scrum ceremonies with Kanban flow for tasks

Notes — Control-Plane (Step Functions + Lambda)

Project. Multi-tenant control-plane to orchestrate tenant lifecycle (onboarding, changes, offboarding) with AWS Step Functions for orchestration, Lambda for compute, EventBridge for scheduling/triggers, and SQS (+DLQ) for decoupling. Strong idempotency (“exactly-once perception”), retries with jitter/backoff, compensations for saga steps, and end-to-end observability.

Orchestration overview

Key decisions & patterns

• Saga orchestration (retries with jitter/backoff, timeouts, circuit-breakers).
• Idempotency: request keys + DynamoDB conditional writes → exactly-once perception.
• Fan-out/fan-in via Parallel; async jobs in SQS (work/retry/DLQ) + outbox.
• Observability: structured logs, RED/USE, X-Ray, canaries; SLO/SLI dashboards.
• Testing: contract tests, local Step Functions, Testcontainers, fault injection.
• Hardening: exponential backoff + jitter, poison-pill quarantine, dedupe, DLQ alarms.

Cold starts vs p95

p95 latency per step — before vs after (ms)

~−30% p95 via batching, payload slimming, fewer network hops and waits.

My role

• Designed state machines, compensations, and idempotency strategies.
• Implemented Lambda handlers, SQS consumers, and inter-service contracts.
• Performance/cost tuning (Provisioned Concurrency, batching, memory tuning).
• Dashboards/alerts and on-call runbooks; cutovers and post-mortems.

Stack

AWS Step Functions, Lambda (Java), EventBridge, SQS, DynamoDB, API Gateway, Terraform, CloudWatch/X-Ray, GitHub Actions; Testcontainers, JUnit/Mockito, contract tests.

Shell Energy

Real-time metering for 2,200,000+ users; multiple meters per user (gas + electricity). A change-frozen XML monolith was wrapped by adapters and gradually strangled to WebFlux microservices on Kubernetes. Observability via Prometheus/Grafana with RPS and hourly/daily/weekly ingestion metrics.

Cutover to microservices25%

drag me

Phase 1 — Assess & Extract

Users

2,200,000+

Meters per user

Multiple (gas + electricity)

Observability

RPS + hourly / daily / weekly ingestion

• Identified bounded contexts around the legacy metering monolith (read-only, XML-only contracts).
• Built adapters / anti-corruption layer: XML over MQ/HTTP, schema contracts, idempotency keys.
• Established observability baseline: Prometheus + Grafana (golden signals, RED/USE), structured logs.
• Prepared data models for WebFlux microservices (CQRS read models for UI).

Highlights

• Event-driven flows with RabbitMQ; outbox + retry/backoff; deduplication.
• Dual-write & shadow traffic with drift detection and safe comparisons.
• WebFlux on Kubernetes (probes, HPA); blue/green & canary via Ingress.
• Latency reduction via I/O tuning & back-pressure; standardized metrics & tracing.

My Role

• Development of WebFlux-based microservices
• Database design and creation
• Data migration / “data pumping” from the monolith
• Deployments and CI/CD pipelines
• Writing unit and integration tests
• Functional discussions with BA/Tech Lead and the team

↑ Back to Projects

SolbegSoft / Helmes — Risk & Asset Protection

Protection software for automotive & property insurance. We adapted the product for a specific region: localization & compliance rules, integrations with regional systems, and configuration-driven feature toggles. In parallel, we migrated selected monolith parts to microservices and hardened QA/stability.

Fewer incidents after QA hardening & adapter isolation (illustrative data).

My Role

• Adapted out-of-the-box features for a specific region (rules, locales, UX).
• Built integration adapters (policy/claims), mapping & validation layers.
• Configuration-driven toggles (feature flags) and rules engine changes.
• Helped extract modules from the monolith to microservices (contracts, endpoints).
• Tests: unit/integration, contract tests for adapters.
• Deployment support and functional scope discussions with BA/Tech Lead.

↑ Back to Projects

Automotive Group — Calendar Management System

Internal Google-style calendar for an automotive group: schedule and manage events with filters, tags, and priorities; speaker profiles (photos + talk details); multi-user views for departments.

Mon

Tue

Wed

Thu

Fri

Sat

Sun

Design Review

Speaker — Helena Mall

UXMed

Design Review

Tag: UX

Priority: Med

Board Meeting

CEO — Ben Smart

ExecHigh

Board Meeting

Tag: Exec

Priority: High

Press Briefing

Speaker — Helena Mall

PRMed

Press Briefing

Tag: PR

Priority: Med

Product Demo

CEO — Ben Smart

Speaker — Helena Mall

LaunchHigh

Product Demo

Tag: Launch

Priority: High

Dept. Standup

InternalLow

Dept. Standup

Tag: Internal

Priority: Low

Partner Sync

Speaker — Helena Mall

PartnerMed

Partner Sync

Tag: Partner

Priority: Med

Town Hall

CEO — Ben Smart

Speaker — Helena Mall

All-handsHigh

Town Hall

Tag: All-hands

Priority: High

TagPrioritySpeaker

Delivery model

• Client provided brand & full design system (Figma) — strict visual parity.
• Waterfall governance, but monthly demos with the client — feedback → iterative scope.
• Sole backend engineer in a 6-person team (1 BE + 5 FE).
• Built from scratch in ~10 months + 2 months production support.

My contributions

• API design (REST), authn/authz (LDAP), audit log.
• Recurring events, time-zones & DST correctness; ICS export.
• Priorities/tags/filters, speaker profiles (photo + talk details).
• CI/CD, Testcontainers, Flyway migrations; performance tuning and caching.

↑ Back to Projects

Contact

Warsaw, Poland · B2B · Contract · Remote/Hybrid/On-site

k30medvedev@gmail.com +48 698 567 174 GitHub LinkedIn

Building scalable, observable cloud systems

Projects

Experience

Backend Software Engineer — Godel Technologies

Backend Software Engineer — Godel Technologies

Backend Software Engineer — SolbegSoft / Helmes Group

Backend Software Engineer — Automotive Group — Calendar Management System

Notes — Control-Plane (Step Functions + Lambda)

Orchestration overview

Key decisions & patterns

Cold starts vs p95

p95 latency per step — before vs after (ms)

My role

Stack

Shell Energy

Highlights

My Role

SolbegSoft / Helmes — Risk & Asset Protection

My Role

Automotive Group — Calendar Management System

Delivery model

My contributions

Contact