Distributed systems
When services, events, and dependencies become hard to reason about.
Ever Cifuentes
Software Architect · Staff Backend Engineer · AI Engineer
14 years · distributed systems · Bogotá, Colombia
Designing resilient software for production at scale.
I design the distributed systems behind streaming platforms, telecommunications networks, and large-scale commerce — the kind where an architectural mistake takes years to surface and years to undo.
- 14+
- Years building software
- 89.6M
- Monthly users reached
- 185+
- Microservices architected
- 6
- Large-scale platforms
Reference architectureevent-driven
sheet01
revA
versionv1.0.0
02
Where this judgment was built
06Large-scale platforms
- 01
Disney Parks
Software Architect & Tech Manager
via Globant ·
- 02
Telecom Argentina
Tech Manager & Software Architect
via Globant ·
- 03
Sky Mexico
Tech Manager & Software Architect
via Globant ·
- 04
Pluto TV / Paramount
Tech Manager & Senior Go Developer
via Globant ·
- 05
Amazon Ring
Senior Software Engineer
via Deviget ·
- 06
Hallmark
Tech Lead & Software Architect
via Globant ·
Six engagements across media, telecommunications, entertainment, and large-scale commerce — architecture, technical leadership, and the assessments that decided what got built. Fourteen years of decisions made under production constraints, and the judgment that came out of them.
03
What I get called for
Not a technology list. The situations where a system stops serving the business.
Cloud architecture
When a platform needs to evolve without stopping the business.
Reliability & observability
When production failures reveal what the system was never designed to show.
AI engineering
When AI needs architectural guardrails, not just an API key.
04
Work
Each engagement is a decision I had to defend, and the reasoning I would give again today.
Engagement 01 / 06
Disney Parks
Software Architect & Tech Manager · via Globant
An administration platform for tens of millions of park visitors, rebuilt in the cloud beside a legacy system nobody was allowed to read.
85+ cloud-native microservices · auth centralized into one auditable layer
Read the architectural decision →
Engagement 02 / 06
Telecom Argentina
Tech Manager & Software Architect · via Globant
An OTT platform that worked every day of the year except the ones that mattered most.
Assessment approved · the next phase of the transformation funded
Read the architectural decision →
Engagement 03 / 06
Sky Mexico
Tech Manager & Software Architect · via Globant
A telecommunications operator moving into television, with a fleet of set-top boxes it wanted to carry a richer channel grid — in FHD and 4K.
Full assessment, target architecture, and migration roadmap delivered — consultancy and pre-sales
Read the architectural decision →
Engagement 04 / 06
Pluto TV — Paramount
Tech Manager & Senior Go Developer · via Globant
89.6M monthly users, 400+ live channels, and a synchronous platform doing asynchronous work.
89.6M monthly active users · 70+ Node.js services migrated to Go
Read the architectural decision →
Engagement 05 / 06
Amazon Ring
Senior Software Engineer · via Deviget
You cannot test a hardware fleet you do not have. So we simulated the fleet, and tested against devices that did not exist yet.
Upcoming devices validated against the software stack before the hardware existed
Read the architectural decision →
Engagement 06 / 06
Hallmark
Tech Lead & Software Architect · via Globant
Thirty-plus Go microservices across commerce, accounts, gifting, and streaming — and the patterns that kept them coherent.
30+ Go microservices · circuit breaking established across teams
Read the architectural decision →
05
How I build
Not slogans. These are the positions I argue for in design reviews, and the ones I have been wrong about often enough to state carefully.
Simplicity is a constraint, not a style
The simplest system that solves the problem is the one that can still be understood at 3am by someone who did not write it. Every abstraction is a loan taken against future comprehension, and the interest is paid during incidents.
Reliability is designed, never added
Circuit breakers, retries, and timeouts bolted onto a finished system only redistribute its failures. Resilience is a property of how services depend on each other, which means it is decided in the design review or not at all.
A system you cannot observe is a system you do not operate
Tracing and metrics are not an operational afterthought — they are how a distributed system becomes legible. If you cannot answer where the latency went without adding code, the architecture already failed at something.
06
Writing
I write about distributed systems, backend architecture, and the parts of engineering that do not fit in a commit message.
BitKode — an engineering field notebook
bitkode.dev
Essays and notes on architecture, Go, and building systems that last.
Read the notebook →
07
Open source
Software I ship, not a portfolio of experiments.
hlab
Go · CLI + TUI · Proxmox
hlab.sh
A Go CLI and full-screen TUI that creates and manages Proxmox VMs and LXC containers. It discovers the cluster over the Proxmox API and drives Terraform and Ansible underneath, so you never hand-write either.
Ships through a Homebrew tap and a curl installer, both verified against release checksums.
08
Facing an architectural decision that will be expensive to get wrong?
Considering Staff+ and Principal engineering roles.