I enjoy building systems where there is more going on than just handling an HTTP request: queues, background workers, containers, retries, failures and services that need to coordinate with each other.

Most of what I work on sits somewhere between backend development and infrastructure. I like figuring out how services should communicate, how to deploy them, how to observe what they are doing and, especially, what happens when something fails.

Lately I've been spending more time on DevSecOps and on running AI workloads locally. I'm interested in the practical side of infrastructure: resource limits, networking, security, performance and the trade-offs behind different deployment decisions.

I also enjoy lower-level topics such as computer architecture and C. They are quite different from infrastructure work, but understanding what happens closer to the hardware helps me reason about systems and performance.

Some of my work lives in private repositories, mainly research software and a mobile application, so the public repositories here are only part of what I've worked on.

I work in English, Spanish and Catalan.

• Local AI infrastructure — exploring what it actually takes to run LLMs on-premises: hardware requirements, GPU and RAM usage, storage, power consumption and the trade-offs between local and cloud inference. I'm building Jaull around this.

• DevOps & distributed systems — experimenting with queue-based architectures, worker scaling, containers, Kubernetes and infrastructure as code.

• DevSecOps — learning how to bring security into the development and deployment workflow through network isolation, scanning, secrets management and safer defaults.

• Homelab — maintaining a small homelab where I can try all of this on real infrastructure, break things, understand why they broke and document what I learn.

GSX Infrastructure · Public
An application stack that I deployed in two different ways: locally with Docker Compose and with Kubernetes managed through Terraform. It includes separate network zones, default-deny NetworkPolicies and monitoring with Prometheus, Alertmanager and Grafana.
Terraform Kubernetes Docker Flask Redis Prometheus Grafana
→ Repository

Scalable and Elastic Ticket Service · Public
A distributed ticket-processing system built around RabbitMQ and PostgreSQL. Producers publish work to a queue, Fargate workers process it, and worker capacity is adjusted according to the queue backlog. The project also covers retries, dead-letter queues, idempotency and concurrent processing, with benchmarks for throughput and latency.
Python AWS ECS Fargate RabbitMQ PostgreSQL Docker
→ Managed-services version · Self-hosted EC2 version

Jaull · Public
A project I'm building to understand whether a local LLM deployment will actually fit and behave well on a given machine before running it. It analyses the available hardware, estimates model memory requirements and validates those predictions against real executions.
Python → Repository

Lipopotamo — scientific data platform · Internal project
A multi-service platform used for scientific data processing around NMR workflows. I've worked on asynchronous workers, external scientific software execution, batch and sample management, report generation, execution history, partial re-runs and versioned data merging. The platform uses Keycloak for authentication and runs as several communicating services with Docker Compose.
Python FastAPI MongoDB React TypeScript Vite Keycloak Docker Compose

Ninjadorks · Public
A search automation tool I built while learning about OSINT and security research. It can build advanced queries for Google and DuckDuckGo, export results, retrieve selected files and analyse collected content with regex or model-assisted processing.
Python Selenium → Repository · Learning labs

Homelab · Public
My small self-hosted lab for learning infrastructure by actually running it. I use it to experiment with networking, containers, monitoring, self-hosted services and automation, while keeping the configuration and documentation in Git.
Python Shell → Homelab · Debian config

_{Infrastructure as code · container orchestration · network segmentation · observability · metrics · alerting · dashboards}

_{Currently going deeper on: Ansible · Helm · Argo CD · Trivy}

_{REST APIs · WebSockets · asynchronous processing · queue-based architectures · background workers}

I'm interested in platform and infrastructure engineering, DevSecOps, distributed systems and on-premises AI. I'm also open to internships where I can keep learning and work on real systems in these areas.