$ whoami
Mathias Rui Lopes_
Lead Full-Stack Engineer · AdTech · Applied AI
I build and run the entire technical backbone of a digital advertising company: the systems that make campaigns run, data flow, and teams operate. From self-hosted LLMs to multi-server infrastructure to the backoffice everyone relies on.
What I do
a/Technical Leadership
lead dev · weekly cycles · linear · code review- Lead developer of the MediaBy tech team since August 2026: I run the development cycle and decide who works on what
- Weekly delivery cycles: large topics broken down into tasks of two days at most, assigned, followed up and reviewed
- Product first: I arbitrate between product work and infrastructure, and keep a single source of truth for the roadmap
- The voice of the tech team to management: written updates twice a week and a report every Friday
b/Full-Stack Development
symfony · node.js · doctrine · rest- Internal backoffice used daily by ~10 people: a complete control layer to manage 2,000+ campaigns and the entire front-end ad-buying platform (Symfony / PHP, Doctrine ORM)
- Machine-learning bricks and API integrations in Node.js
- MCP (Model Context Protocol) APIs connecting LLMs to internal business data
- Ingestion pipelines for DSP campaign data
c/Infrastructure & DevOps
docker · nginx · redis · linux- Administration of 5 physical and cloud servers, plus multiple Docker services (SSH, Apache2, Nginx, VHOST)
- Dedicated vGPU cloud server deployed and operated for AI workloads
- Monitoring (Netdata), real-time (Mercure), caching (Redis, OPCache)
- Docker-based recovery delivering near-100% uptime, with incidents typically resolved in under 20 minutes
- Network security, firewalls, GitHub management
d/Applied AI
ollama · mcp · llm · production- Proprietary AI built on a sovereign and self-hosted LLM stack deployed and operated in production
- Optimizes campaign performance across DSPs and turns complex DSP reporting into clear, actionable insight
- MCP protocol implementation: AI ↔ internal database interaction
- Model testing, qualification and benchmarking
- Staying ahead of the LLM ecosystem and its business applicability
e/Trafficking & AdOps
dsp · pixels · server-side tagging- Technical co-management of the trafficking team, DSP expertise across major platforms
- Pixel & tag implementation, client-side and server-side
- Data-integrity validation whatever the client stack (web, app, custom)
- Direct operations coverage during team absences
Experience
Innovation Director & Lead Developer
MediaBy · digital advertising agency
Sole senior technical profile of the company: I own the entire technical scope (full-stack development, infrastructure, applied AI and cross-functional IT) and act as the technical interface between business teams and the systems they depend on.
- Build and maintain the internal backoffice used daily by ~10 people to manage 2,000+ campaigns and the front-end ad-buying platform (Symfony, Doctrine, Twig, ChartJS)
- Build ML components and API connectors in Node.js, including ingestion of Adform DSP data via S3 buckets
- Operate 5 physical and cloud servers plus multiple Docker services, including a dedicated vGPU server for AI
- Containerise critical services with Docker for near-100% uptime and sub-20-minute incident recovery
- Run self-hosted LLMs in production, wired to internal data through MCP APIs
- Use them to optimize campaign performance across DSPs and make complex DSP reporting easy to read
- Qualify models and AI tooling against concrete business use-cases before adoption
- Lead the development team since August 2026: weekly cycles, task breakdown and assignment, code review, production releases
- Report on delivery to management every week and turn their priorities into a clear technical plan
- Co-manage the trafficking team and translate business needs from CDPs into technical solutions
- Design architectures from single-server setups to multi-server deployments
← earlier roles: 10+ years across the AdTech industry · full history on LinkedIn
Stack
Contact
Want to talk engineering, AdTech or running AI in production? My inbox is open.