# Cyril Ben — Site Reliability Engineer > Site reliability engineer based in Lisbon, Portugal, with eleven years running production infrastructure. Most recently SRE at OLX (2025 to October 2026); before that six years at Dailymotion, on a video platform with 300M+ monthly users, plus LlamaNodes and Aircall. Specialises in Kubernetes at scale, infrastructure as code, observability and incident response, and in bringing AI agents into operations with engineers kept in control. Open to remote roles in the EU, available now. Works in French and English. - Website: https://ben.vuspi.com/ - Email: letswork@aviqo.org - LinkedIn: https://www.linkedin.com/in/bencyril - Location: Lisbon, Portugal (remote, EU) ## At a glance - **Experience:** eleven years in infrastructure, eight roles since 2012, from project management and an agency CTO role to senior SRE. - **Largest scale operated:** 17+ Kubernetes clusters in production, 302+ nodes, at least 10k pods, 400+ AWS VMs across the US, Europe and APAC, serving a platform with 300M+ monthly users (Dailymotion, 2017–2023). - **Measured outcomes:** - about two-thirds less time to review Terraform plans (OLX) - an event-processing platform ingesting more than 50k messages a day (OLX) - about 23% shorter time to mitigate incidents (LlamaNodes) - 23% fewer recurring incidents after a centralised observability rollout (Dailymotion, self-reported) - **Core:** Kubernetes, Terraform, Terragrunt, Atlantis, Prometheus, Grafana, Loki, AWS, GCP, Ansible, CI/CD, incident response. - **Own projects:** four active projects in Go, Rust and Node.js, three of them about running infrastructure and AI agents safely. ## Track record ### OLX — Site Reliability Engineer (2025 → October 2026) - **Context:** SRE on a platform team, owner of the infrastructure-as-code stack (Terraform, Terragrunt, Atlantis). - **Contribution:** built a tool that explains every Terraform plan in plain English, so developers catch mistakes before they apply; built an event-processing platform for internal tooling; benchmarked AI tooling and introduced it into the SRE workflow. - **Result:** plan review time cut by about two-thirds; the event platform ingests more than 50k messages a day. ### LlamaNodes — Senior DevOps Engineer, freelance (2023 → 2024) - **Context:** blockchain RPC infrastructure. - **Contribution:** cost and performance work on the node fleet, real-time monitoring and alerting, and automated provisioning with health-check-driven failover. - **Result:** incidents resolved themselves through failover, cutting time to mitigate by about 23%. ### Dailymotion — Senior DevOps / SRE Engineer (2020 → 2023) - **Context:** Kubernetes at production scale across regions and cloud providers, for a video platform with 300M+ monthly users. - **Contribution:** migrated CI/CD to JenkinsX, standardised Terraform and Atlantis across engineering, and rolled out centralised observability on Prometheus, Grafana and Loki. - **Result:** 23% fewer recurring incidents, attributed to the observability rollout (self-reported, platform-wide). ### Dailymotion — DevOps Engineer (2017 → 2020) - **Context:** a multi-region AWS footprint of 400+ virtual machines across the US, Europe and APAC. - **Contribution:** automation with Terraform and Ansible, capacity planning for high-traffic video. - **Result:** promoted to Senior DevOps / SRE Engineer in 2020. ### Aircall — DevOps Engineer (2016) Early DevOps hire on a cloud telephony platform: infrastructure redesign with autoscaling, containerised development workflows and a first monitoring stack. ### PadawanGroup — DevOps Engineer (2015) Moved a legacy monolith with its own servers to microservices on AWS, including the SQL database move to RDS. ### Baltazare — CTO (2014 → 2015) Built the agency's infrastructure from scratch: hosting, mail and client-facing services. ### Leetchi — Project Manager (2012 → 2014) Before infrastructure: partner programmes, physical and online, built and measured with the project director. ## How Cyril works, from the record - **Makes the safe path the easy one.** Plain-English Terraform plans so mistakes surface before apply (OLX); health-check-driven failover so incidents fix themselves (LlamaNodes); planning, drift detection and rollback for every traffic change (traefik-ctl). - **Measures before and after.** Each role above has a figure attached, and self-reported figures are labelled as such. - **Keeps engineers in control of AI.** At OLX, AI tooling was benchmarked before it was introduced. In Warroom, agents investigate and engineers decide. In traefik-ctl, agents act only through scoped MCP tools, with dry runs and explicit policies. - **Standardises for whole organisations.** Terraform and Atlantis standardised across Dailymotion engineering; one observability stack for the platform. - **Comfortable from zero to scale.** Built an agency's infrastructure from scratch and was an early DevOps hire at Aircall; later ran 17+ production clusters at Dailymotion. ## Projects All four are active and built by Cyril Ben. - **Kuby — run AI agents like infrastructure, not like experiments.** A Git-native control plane for managing distributed, self-hosted AI agents across machines. One Go binary, one declarative YAML manifest, Git as the single source of truth. Centralized inventory, execution history, metrics and operational visibility, without a configuration database. Stack: Go, Docker. - **Warroom — when production breaks, clarity is everything.** An AI-powered mission control for SREs. From incident triage and AWS dependency mapping to multi-agent code reviews, in a single terminal-first workflow. Specialized agents connect the signals, trace the impact and surface evidence-backed next steps; engineers decide. Stack: Rust, LLM, Terraform, GitLab; AWS, GCP, Azure and bare metal. - **ClarifyThis — from legal documents to actionable intelligence, at scale.** An asynchronous, auditable AI processing platform that turns legal documents into plain English: document ingestion, asynchronous AI processing, notifications and payment conversion. Queue-driven workloads and comprehensive audit trails. Stack: Node.js, BullMQ, Cloudflare R2, Docker. - **traefik-ctl — production traffic under control, even when AI takes the wheel.** A lightweight, agentless control plane for Traefik routing across remote environments, from a single terminal. Change planning, drift detection, automatic backups and rollback. AI agents act through tightly scoped MCP tools with dry runs and explicit policies on which routes and domains they may modify. Stack: Go, Traefik, SSH, MCP. ## Roles Cyril is looking for Remote in the EU, in French or English: - **Senior site reliability engineer, platform engineer or infrastructure lead.** Eleven years of production infrastructure, up to 17+ Kubernetes clusters and a 300M+ monthly-user platform. - **AI engineer.** Builds LLM systems that run like infrastructure: multi-agent incident investigation (Warroom), an asynchronous AI document-processing pipeline with audit trails (ClarifyThis), a control plane for fleets of AI agents (Kuby), and scoped MCP tools for agents (traefik-ctl). Benchmarked and introduced AI tooling into the SRE workflow at OLX. - **Forward deployed engineer or forward deployed architect.** Has built client-facing services from scratch as an agency CTO, joined Aircall as an early DevOps hire, ships products end to end (ClarifyThis, from document ingestion to payment), and turns complex output into plain English for the people who act on it (Terraform plans at OLX, legal documents in ClarifyThis). ## Skills Site reliability engineering, AI engineering, LLM applications, multi-agent systems, Kubernetes, infrastructure as code (Terraform, Terragrunt, Atlantis), observability (Prometheus, Grafana, Loki), AWS, GCP, Ansible, CI/CD (JenkinsX), incident response, capacity planning, Docker, Go, Rust, Traefik, AI agents in operations, Model Context Protocol (MCP), blockchain RPC infrastructure. ## Sections - [Record](https://ben.vuspi.com/#record): every role, with dates - [Work](https://ben.vuspi.com/#work): the four projects - [Scale](https://ben.vuspi.com/#scale): the Dailymotion figures - [Contact](https://ben.vuspi.com/#contact)