Elliot Morales Solé
VP of Engineering · Real-time video, Kubernetes and AI-assisted engineering
Almería, Spain · 100% remote, available to travel
Download CV · PDF, 2 pagesTwelve years building and running high-traffic platforms, the last ten on the same live-streaming platform, where I have gone from PHP developer to VP of Engineering. I design and write the whole infrastructure — on-premises Kubernetes, BGP networking, databases, CI/CD — along with the video pipeline and the PCI payments software. Before that I was co-founder and CTO of a hosting and systems company. I still write code every day in Go, PHP, TypeScript and Python.
Some things I have built
a-stream — fully DRM-protected, ultra-low-latency live video 2026
A multi-tenant streaming platform delivering 0.3–0.8 s glass-to-glass latency with full DRM on every stream — not one or the other. I am its architect and lead developer. All three DRM systems are supported: Widevine, PlayReady and FairPlay. Content is encrypted once at the origin, so the bytes are identical on every edge and a viewer can fail over between them without renegotiating the session. WHIP/RTMP/SRT ingest, zero-transcode origin→edge cascade, recording and clips. It went from pilot to 25% of production streams in three weeks.
I am currently rewriting the edge: 100% Go, with a WebSocket transport to the viewer replacing WebRTC. The reason is that the DRM player had already stopped using the browser's video decoder: it taps the encoded frames — still encrypted — repackages them into CMAF in WebAssembly with their subsample map, and hands them to MSE. Decryption happens where it must, inside the browser's CDM through EME; that is what makes it real DRM. WebRTC had become an expensive pipe for bytes we already knew how to handle. On the bench, time to first frame drops from seconds to milliseconds on the affected path. In progress, running alongside the current version.
Widevine license server 2026
In-house server written in Go, wrapping the official C++ SDK through cgo. Certified Widevine Implementation Partner with Google — Modular DRM & License SDK, 2025. I have also built PlayReady and FairPlay license servers — all three DRM systems on the market.
hls-edge 2020–2021
An HLS edge server written entirely in Go, and single-handedly: active-active high availability on etcd, gRPC with mutual TLS, distributed cache, an ffmpeg-based remuxer with a watchdog, early LL-HLS, HTTP/3 and cross-cluster replication.
Percona 5.7 to 8.4 in two months 2026
An in-place upgrade was impossible: an earlier migration had left the date columns in a format 8.4 no longer accepts. It was solved by rebuilding the database logically — multi-threaded dump and restore — with GTID replication across versions and promotion of a replica to primary, over six servers and with a rollback path ready.
PCI DSS SAQ-D card gateway since 2021
Designed and maintained by me, with a current audit renewed three times a year. Over 2,000 transactions a day across nine integrated payment providers, with 3D Secure, recurring subscriptions and fraud control.
On-premises Kubernetes, from the metal to the player
Five production k3s clusters of 10 to 20 nodes each: Cilium with BGP and BFD, replicated route reflectors and ECMP for the load balancers, HAProxy ingress, Ceph RBD, Longhorn and NFS, on Proxmox deployed with Terraform and Ansible.
DevEnv 2026
A Go platform that spins up a complete development and pre-production environment per developer on Kubernetes, with more than 15 services per namespace, reachable over REST and MCP, with data seeding and multi-tenant isolation.
From Jenkins to GitHub Actions
CI migrated to ephemeral runners on Kubernetes: the backend build went from 17 minutes to 41 seconds warm, with more than 12,000 tests spread across seven parallel jobs.
An engineering organisation assisted by AI agents 2026
I designed the team's workflow on Claude Code: per-domain agents with explicit roles and contracts, automated code review, an orchestrator that takes a support bug through to a pull request, spec-driven work and cost telemetry.
Career
Teamsquad
Nov 2022 – presentVP of Engineering · remote
End-to-end owner of a high-traffic live-streaming platform with payments, real-time chat and WebRTC/HLS video: from the hardware in five data centres to the player. A team of up to four people and an engineering budget of around $300k a year.
VTS Media
Apr 2016 – Sep 2022PHP Developer → DevOps → Senior DevOps · Barcelona
The same platform, before the company moved to the United States. Six full production migrations, from Docker Cloud to Rancher and then to Kubernetes, and between our own data centres. A streaming system built entirely in house, and a PHP backend evolved to DDD/CQRS over RabbitMQ.
Brutal Systems
Feb 2014 – Nov 2016CTO and co-founder · Barcelona
A systems, cloud and DevOps company: colocation, dedicated servers, VPS, domains and hosting. Built complete infrastructures from scratch and handled DDoS mitigation with BGP. Exited through the acquisition of my shares.
Universitat Politècnica de Catalunya
2012Researcher, WikiSat project · Castelldefels
Designed the convergent-divergent nozzle for the second-stage rocket engine of the launcher, validated in static laboratory firings.
What I work with
Open source
- Home Assistant — fix to the Alexa integration, merged (2021)
- FastNetMon — next-hop on ExaBGP withdrawals, merged (2016)
- LiveKit — in-house fork in production; upstream contribution in preparation