I am a Systems Engineer from Medellín, Colombia. I like working close to the system: Linux, clusters, networks, virtualization, observability, and the software that sits on top of them.
At the Scientific Computing Center APOLO, I worked across HPC operations, Linux administration, Slurm, FreeIPA, Ansible, Prometheus, Grafana, and virtualization. I carried out the migration of seven machines from VMware ESXi to Proxmox with zero downtime. Under equivalent production workloads, average node load moved from 55% to 35%.
I was also the HPC Infrastructure Lead for the team that won the ASC26 Student Supercomputer Challenge in Wuxi, China, where we optimized ICON inference for weather and climate forecasting.
My GitHub is a record of the things I learn by building: low-level networking in C, fault-tolerant middleware in C++, cloud-deployed ML services, CI/CD pipelines, and GPU-backed experiments.
Infrastructure & HPC
Software & distributed systems
Cloud, delivery & AI
| Project | What I built | Main tools |
|---|---|---|
| High-availability distributed middleware | Leader–Follower replication with persistence, failure handling, Point-to-Point queues, and Publish–Subscribe topics. | C++ · gRPC · Protocol Buffers · AWS |
| DHCP-Project | A DHCP client and server implemented from the protocol level for a NAT network. | C · UDP sockets · Linux networking |
| GRID-EAFIT energy predictions | Energy-consumption and solar-production prediction services packaged as Dockerized backend/frontend services. | TensorFlow · Flask · React · Docker Compose · AWS EC2 |
| Neural machine translation | English-to-Spanish sequence-to-sequence training with attention, GPU execution, and Slurm job configuration. | TensorFlow/Keras · CUDA/cuDNN · Slurm |
See the rest of my coursework and experiments on my repositories page, including Hadoop MapReduce, computer vision, concurrency, numerical methods, and CI/CD automation.