Backend services
APIs and services with clear contracts, database transactions, and tests for failure cases.
Backend engineer · Porto, Portugal
Backend and platform engineering.
I’m Gabriel, a backend engineer at Mercedes-Benz.io. I work mainly with Java, Kotlin, Spring Boot, AWS, Kafka, SQS, PostgreSQL, and Kubernetes.
Expertise
What I work with.
I build and run backend services, event-driven integrations, and the infrastructure around them.
Messaging and async work
Kafka and SQS integrations, idempotent consumers, outbox patterns, retries, and replay.
Platform and operations
AWS and Kubernetes deployments, GitOps workflows, metrics, logs, and automated security checks.
Mercedes-Benz.io
Backend engineering for digital products.
My work covers backend services, event-driven integrations, and platform tooling used to build and operate them.
Selected work
Webhook Relay.
A reference project showing how I approach durable acceptance, at-least-once delivery, retries, dead letters, and replay.
Open source · Kotlinv0.1.0
Webhook Relay
I built this Kotlin and Spring Boot project to make webhook failure handling concrete. It uses PostgreSQL for persistence, Redpanda for messaging, and Testcontainers for integration tests.
- Transactional PostgreSQL outbox with Kafka-compatible Redpanda transport
- HMAC-signed, at-least-once delivery with idempotent worker claims
- Exponential full-jitter retries, dead-letter persistence, and replay APIs
- OpenAPI docs, structured logs, health checks, Prometheus metrics, and containerized tests
Open source
Open-source work.
A maintained fork and a merged upstream contribution.
graphql-sqs-subscriptions · maintained fork
I upgraded this fork to AWS SDK v3 and maintain it for a project at Mercedes-Benz.io.
httpexpect · merged pull request #337
Contributed opt-in request and response controls to the project’s default formatter, including end-to-end test coverage.
Writing
Backend engineering notes.
Notes about reliability patterns I implemented while building the public projects on this site.
Reliable webhook delivery: designing for the failures between systems
Notes from building Webhook Relay: transactional outboxes, at-least-once delivery, idempotency, retries, dead letters, replay, and request signing.