About Me

Backend Software Engineer

4+ years building backend services and microservices architecture, specialized in Go. Passionate about distributed systems, clean architecture, and performance optimization.

$cat resume.pdf

Why I Built This Blog

I started RuntimeRants out of curiosity and a need to document what I learn.

Most tutorials tell you what to do, but rarely explain why it works that way. I wanted a space to share the deep dives I do when learning — the real engineering notes, not the polished marketing content.

What you'll find here:

Technical deep dives into backend systems
CI/CD pipelines, deployment strategies, and DevOps patterns
Low-level explorations: sockets, syscalls, concurrency
Comparisons with code, benchmarks, and honest opinions
Notes from books I'm reading (DDIA, system design, etc.)

What I Do

I focus on Backend & Systems Engineering — building scalable, reliable services that handle real-world complexity.

Backend Development & APIs

Backend development, API design, and building production services.

Distributed Systems & Data Pipelines

Distributed systems, data pipelines, messaging/streaming, and reliability.

Architecture & System Design

Architecture, system design, scalability patterns, and real-world trade-offs.

Experience

Backend Software Engineer (Golang)

Dribba

Oct 2024 – PresentAndorra.com · CityXerpa

Designing, developing, and maintaining backend microservices in Golang within distributed, production-grade environments.

Own the full product lifecycle of two microservices from scratch — API research, design collaboration, GraphQL contract with frontend, DB schema, architecture, caching strategy, and deployment — reporting directly to the CTO with near-total autonomy
Designed the multi-vertical global search on Elasticsearch: a separate index per vertical instead of one shared index, federated queries via _msearch, and self-contained card-level documents so the UI renders a result without a second trip to the database (⟨N⟩ documents across ⟨N⟩ verticals, p95 ⟨latency⟩)
Migrated an ENUM column on a live production MySQL table with ALGORITHM=INSTANT, then moved the domain to a lookup table — no downtime, no table rebuild (⟨N⟩-row table)
Built the billing layer for bus reservations: IGI tax and commission chained entirely in cents so no floating-point drift accumulates across the calculation; investigated and resolved a billing discrepancy with an external provider
Integrated the full bus booking cycle with the Movelia provider: purchase, cancellation, and ticket PDF generation
Unified the GraphQL schema for direct and multi-segment routes so the frontend consumes a single shape regardless of how the trip is structured underneath
Integrated Redsys as the card payment gateway with transactional safety — states, async callbacks, retries, duplicate prevention — so a failed payment never yields a confirmed booking
Designed the Redis caching strategy, hotel image deduplication, and the cancellation feature end to end
Designed services with hexagonal architecture, clean code, and separation of concerns across GraphQL, gRPC, and REST; deployed on Kubernetes (GKE) with GitLab CI/CD on GCP

GoMySQLElasticsearchRedisgRPCGraphQLKubernetesGCP

System Design

I work through large-scale designs end to end — real-time feeds, ticketing, ride-hailing, timeline fan-out — then build the pieces I find most interesting as small, runnable systems.

Real-time fan-outSharded WebSocket hub, per-connection write pump, non-blocking send so one slow client never stalls a broadcast Distributed rate limitingRedis token bucket — one shared limit across N nodes, guaranteed by an atomic Lua script Job schedulingMin-heap priority queue with sync.Cond, so idle workers sleep instead of spinning Circuit breakerTrips on sustained failures and fails fast instead of piling requests onto a dying dependency SingleflightCollapses duplicate in-flight calls for the same key into one, so a cache miss stampede hits the backend once Bounded parallel fetcherrgroup with a concurrency limit — fan out, cap the blast radius, cancel the rest on first error Fan-out, first-response-winsRace N replicas and take the first answer back, so tail latency stops dictating the response time

The other half is storage internals: LSM trees and partition design in Cassandra, B+ trees and locking in MySQL, write-ahead logs and change data capture.

How I Practice

System design exercises — I work Hello-Interview-style designs end to end (Uber, ticketing, real-time feeds, timeline fan-out), and when something nags me I pull the thread down to the internals — that's how I ended up reading about Cassandra and RabbitMQ
Advent of Code (opens in a new tab) — every year
LeetCode (opens in a new tab) — to keep the algorithmic muscle sharp
Reading — Concurrency in Go (Cox-Buday), finished and re-read; Designing Data-Intensive Applications (Kleppmann), halfway through

Education

Software Engineering Degree

Universitat Rovira i Virgili

Sept 2020 – June 2024 · Tarragona, Spain

Tech Stack

The tools I use regularly for backend development and infrastructure.

stack — zsh

joel@runtimerants ~/stack % tree

Tech Stack

├─Languages

│ ├─Go

│ ├─Java

│ ├─C

│ ├─Python

│ └─JavaScript

├─Backend & APIs

│ ├─Spring Boot

│ ├─gRPC

│ ├─GraphQL

│ ├─REST

│ ├─WebSockets

│ ├─Webhooks

│ └─Sockets

├─Databases

│ ├─PostgreSQL

│ ├─MySQL

│ ├─Redis

│ ├─Elasticsearch

│ └─Cassandra

├─Messaging & Streaming

│ ├─Kafka

│ ├─RabbitMQ

│ └─Pub/Sub

├─Architecture

│ ├─Microservices

│ ├─Event-driven

│ └─Hexagonal

├─Cloud & Containers

│ ├─GCP

│ ├─Cloud Run

│ ├─Kubernetes

│ └─Docker

└─DevOps & Tooling

├─GitLab CI/CD

├─Git

├─Linux

└─Bash

joel@runtimerants ~/stack %▋

Contact

GitHub LinkedIn Email Instagram

Feel free to reach out if you want to discuss something technical, collaborate on a project, or just say hi.

Setup / Uses

setup — zsh

editorsNeoVim · VS Code · IntelliJ IDEA

terminalmacOS Terminal · Oh My Zsh

vcsGitHub · GitLab

containersDocker Desktop · Minikube

api_toolsPostman · Swagger / OpenAPI

databaseTablePlus

networkingngrok · Tailscale · NordVPN

notesObsidian · Excalidraw

Hardware

LaptopMacBook Pro (M4 Pro) · 24 GB · 512 GB

MonitorBenQ Mobiuz 27" 2K · 175 Hz

Desktopi5 14th Gen · RTX 4070 Super · 32 GB · 2 TB SSD

KeyboardEpomaker F75 "Vanilla Ice"

MouseLogitech G502

HeadsetLogitech Gaming

Side Projects

go-sharded-ws-hubWebSocket gateway with a sharded hub. Fall behind and you get dropped — a slow consumer must never stall the broadcast.go-redis-token-bucketRedis token bucket with atomic Lua. One shared limit across N nodes, no drift.go-priority-schedulerMin-heap priority scheduler with sync.Cond. Idle workers sleep instead of burning CPU.CrispLite1:1 real-time chat: WebSockets, Redis Pub/Sub, PostgreSQL, hexagonal architecture.DDD-EcommerceE-commerce backend with real bounded contexts, aggregates that defend their own invariants, and domain events.my-http-serverHTTP server written from scratch over raw TCP. The fastest way to stop treating net/http as magic.

This Blog

framework: Next.js + Nextra
styling: Custom CSS
hosting: Vercel
analytics: none (privacy first)