AI Native Netherlands

Engineering with AI Agents: From Multi-Agent Builds to Agent-Ready Codebases

Thu, Sep 3 · 6:00 PM CEST

Buro de Pijp, 72 Eerste Jan Steenstraat 1072 NN, Amsterdam, NL

We are thrilled to welcome you to our 11th edition of the AI Native Netherlands, hosted at Buro in De Pijp, Amsterdam.

This is our first meetup after the summer holidays, and we're excited to have you back with the AI Native Netherlands community.

A massive thank you to Cloudflare for sponsoring this evening and taking care of the food, drinks, and venues, and for supporting the Dutch AI engineering community.

This edition is about doing engineering work with agents — from orchestrating multi-agent workflows that build new systems, to getting AI agents to safely navigate the messy, legacy codebases you already have.

We'll cover:

• Putting coding agents to work on real production systems — and where they deliver versus where they produce confident nonsense
• Orchestrating multi-agent workflows that handle the heavy infrastructure lifting: a deployment pipeline cut from 4 hours to 4 minutes, and a distributed, edge-native sportsbook
• Getting agents oriented in unfamiliar, legacy code — and making a codebase navigable enough to trust them in it
• Turning the tech-debt backlog you keep postponing into work agents can actually chip away at, without resorting to a rewrite
• The "battle scars": the cognitive load of managing code you didn't write, and fixing hallucinated API integrations
• The habits that make AI-assisted engineering safe, and what "value" really means for augmented engineering

Speaker 1: Raul Tavares (Cloudflare)
Raul Tavares is a Senior Solutions Engineer at Cloudflare. With previous cloud architecture experience at Google and AWS, his current focus is the intersection of edge computing and AI-augmented workflows. Passionate about practical execution over pure theory, Raul helps developers leverage modern AI tooling to build faster, more resilient distributed systems.

Talk: From Building to Orchestrating: Engineering with Multi-Agent AI
He'll share how he uses AI to manage architectural complexity and accelerate development, prototyping, and time-to-market. He'll demonstrate how CodeGen AI handles heavy infrastructure lifting by walking through two real-world projects he recently built: an agentic workflow that cuts website deployment from 4 hours to 4 minutes, and a distributed edge-native sportsbook. He'll bring customer stories, dive into the "battle scars", the cognitive load of managing code you didn't write, fixing hallucinated API integrations, and defining what value means for augmented engineering with AI tools.

Speaker 2: Michiel van der Ros (TomTom)
Michiel van der Ros is a Staff Engineer at TomTom and tech lead of its web development team. Over two decades he's worked on everything from interactive documentaries and games to large corporate websites. Frontend is his passion, but he's just as at home in cloud architecture and AI-assisted development. He's a seasoned conference speaker, always up for a good discussion, a bit of fire juggling, or a game of table football.

Talk: Rerouting: From Tech Debt You've Been Putting Off to a Codebase Agents Can Actually Navigate
Every team has a list of cleanups it never gets to: the module everyone tiptoes around, the half-finished migration, the refactor that always slips to next quarter. The old fix was often a rewrite — which usually made things worse.

There's a better route now. Coding agents can chip away at that backlog for you, but only if they can find their way through your codebase. Aim one at a messy repo blind and you get confident nonsense; give it the right map and it delivers.

He'll show how he put agents to work on an existing production frontend: how he gets them oriented in unfamiliar code, the habits that make them safe to trust, and where it still goes wrong.

Agenda:

• 18:00 — Arrival, food & drinks
• 18:30 — Talk #1 | Raul Tavares
• 19:15 — Talk #2 | Michiel van der Ros
• 20:00 — Networking & more drinks
• 21:00 — Wrapping up

What to bring:
Just curiosity and questions. If you're working on multi-agent workflows, AI-assisted development, or wrestling with legacy code and tech debt in the age of agents, we'd love to hear your questions.

Who is this for:
Platform engineers, AI/ML engineers, SREs, architects, frontend and full-stack engineers, and engineering leaders focused on building and validating reliable AI systems in production.

Where to find us:
Buro, Eerste Jan Steenstraat 72, 1072 NN Amsterdam. In the heart of De Pijp, a short walk from tram stops on Ferdinand Bolstraat.

The AI Coding Adoption Ladder & The MCP Wire

Tue, Sep 29 · 6:00 PM CEST

Studytube, Danzigerkade 17, Amsterdam, NH, NL

We are thrilled to welcome you to our 12th edition of AI Native Netherlands, hosted at the Studytube headquarters in Amsterdam.

In this edition, Pini takes the organisational view of AI coding adoption - how it progresses across an engineering org and the order things tend to break as it scales. Aleksei goes hands-on into the tooling - what happens when you build an MCP server the way the docs suggest, where it falls apart, and the pattern that replaces it.

A massive thank you to our host, Studytube, for their generosity in providing the venue and drinks for this evening and for supporting the Dutch AI engineering community.

We'll cover:

• How AI coding adoption progresses across an engineering org, from first licences toward a mostly automated delivery process
• The order things usually break as adoption scales: coordination, review, testing, context, security boundaries, ownership, CI/CD, and product flow
• Why faster individual developers don't automatically mean faster team delivery
• The platform and infrastructure foundations for scaling AI-generated code safely
• Where traditional MCP tools fit, and where they hit their limits
• Giving a model an API plus a sandboxed code environment instead of writing a new tool for every use case
• Running model-generated code safely: sandboxing, module allowlists, CPU and memory limits, static analysis, and audit logs

Speaker 1: Pini Reznik (re:cinq)
Pini Reznik is a seasoned technology leader with extensive experience in the software industry. He is the CEO and Co-Founder of re:cinq, experts in Cloud Native and AI Native engineering. Former software engineer, Co-Founder of two successful consulting businesses and an O’Reilly author.

Talk: The AI Adoption Ladder: From First Licences to the Dark Factory
AI coding tools make individual developers faster, but many teams discover that delivery does not improve at the same rate. PR volume rises, review queues grow, tests become less convincing, context gets messy, and eventually product and platform bottlenecks become visible. AI did not remove the constraint; it moved it.

This talk introduces the AI Coding Adoption Ladder: a practical model for how engineering teams move from first licences to structured adoption, agentic workflows, and eventually software factory patterns. We will look at the order things usually break - coordination, review, testing, context, security boundaries, ownership, CI/CD, and product flow - and what software and DevOps engineers need to fix before AI-generated code can safely scale.

Speaker 2: Aleksei Moiseev (Studytube)
Aleksei is a software architect and engineer with 20+ years of experience building production systems, from hands-on development to large-scale architecture and platform engineering.

Today, he focuses on agentic AI, designing software factories where LLMs and autonomous agents perform real engineering work. His work spans self-hosted AI infrastructure, multi-agent systems, and the architectural patterns needed to run them reliably in production.

He builds and operates the systems he writes about, with a focus on practical, resilient engineering over hype.

Talk: The MCP Wire
Build an MCP server the way the docs tell you to — one API endpoint, one tool — and it works. Then you ask it to create 1,000 users and wait 25 minutes while it makes 1,000 separate calls. Multi-step workflows turn into long chains. Every new use case is another tool to write, test, and maintain. Adding "bulk" tools just moves the problem around.

Aleksei, an AI Solutions Architect at Studytube, walks through building exactly this server for Studytube's user and team management API, hitting that wall, and arriving at a different pattern: give the model an API and a sandboxed Python environment, and let it write and run code for the task in front of it. Ten tool calls become one short loop, executed once — about two seconds instead of fifteen.

The talk covers where traditional MCP tools still fit (simple, atomic, read-only operations), where code execution takes over (batch jobs, conditional logic, multi-step workflows), and how to run model-generated code safely: sandboxing, module allowlists, CPU and memory limits, static analysis, and full audit logs.

A hands-on talk for anyone building with MCP, or about to.

Agenda:
18:00 — Arrival, food & drinks
18:45 — Talk #1 | Pini Reznik
19:30 — Talk #2 | Aleksei Moiseev
20:15 — Networking & more drinks
21:00 — Wrapping up

What to bring: Just curiosity and your own questions. If you're wrestling with agentic workflows, CI/CD under AI-generated code, or security boundaries as adoption scales — or if you're building with MCP and hitting its limits — bring those.

Who is this for: Platform engineers, AI/ML engineers, SREs, architects, and engineering leaders working on reliable AI systems in production — and anyone building with MCP, or about to.

Where to find us: Studytube, Danzigerkade 17, 1013 AP Amsterdam — Google Maps Link

Right-Sizing AI: Compute, Cost, and the Right Tool for the Job

Tue, Oct 27 · 6:00 PM CET

AWS Amsterdam, Mr.Treublaan 7, 1097 DP Amsterdam, Netherlands, Amsterdam, NL

AI Native Netherlands is back! We'll be hosting our 13th edition at the AWS office.

Before anything else - thank you. This community has grown faster than we ever imagined, and it's been great to watch the connections be made, and the conversations getting better at every edition. The AI landscape keeps moving fast, and there's no better way to keep up than in a room full of people working through it alongside you.

We've got two amazing speakers lined up and a great venue for the evening — we'd love to see you there.

A huge thank you to our friends at AWS for hosting us at their Amsterdam office. Food and drinks will be provided!

We'll cover:

• When to run a task on a CPU versus a GPU — and what that choice does to cost.
• Using fine-tuned small models (SLMs) for classifying, routing, embedding, and reranking.
• Building a real CPU-first agent on Amazon EKS that calls an LLM only when needed.
• Measuring price/performance in practice — and the design assumptions that turned out to be wrong.

Speaker 1: Christian Melendez (AWS)
Christian Melendez is a Principal Specialist Solutions Architect at AWS. He helps the region's largest enterprises build efficient, resilient AI and cloud-native workloads on Kubernetes, with a focus on compute efficiency, cost optimisation, and autoscaling at scale. Author of Kubernetes Autoscaling and creator of Karpenter Blueprints, a best-practices repository that grew Karpenter adoption 16.5x across EMEA, he also built Slemify, an open-source framework for fine-tuning and serving Small Language Models on Kubernetes. A regular speaker at AWS re:Invent, KCDs / CNDs, ContainerDays, and others, Christian focuses on making infrastructure simple, observable, and cost-effective at scale.

Talk: The Right Tool for the Right Token: When to Reach for a CPU or a GPU
While hard reasoning problems require an LLM and a GPU, other tasks can be solved with simpler means. Classifying, routing, embedding, and reranking are structured, high-frequency jobs that a fine-tuned small model (SLM) can handle on CPU at predictable cost. In this session, we will demo a pipeline where each task runs on the best fitting infrastructure. This will include a real CPU-first agent on Amazon EKS running an SLM, and calling an LLM when needed. We will share our learnings from building this pipeline, including how we measured price/performance, its limitations, and which of our design assumptions we found to be wrong. You will leave with confidence to choose the right tool, and an open-source reference architecture to validate your assumptions.

Speaker 2: William Rizzo (Mirantis)
William Rizzo is Global Field CTO at Mirantis, where he helps organisations design, build, and run platform engineering, edge, and AI infrastructure initiatives. His career spans engineering, pre-sales, product ownership, and consulting across high-performance computing, storage, and distributed systems. A CNCF and Linkerd Ambassador and a Kairos maintainer, he's a regular speaker at KubeCon on platform engineering and building resilient internal developer platforms.

Talk: To be confirmed.

Agenda:
18:00 — Arrival, food & drinks
18:45 — Talk #1 | Christian Melendez (AWS)
19:30 — Talk #2 | William Rizzo (Mirantis)
20:15 — Open conversation, networking & more drinks
21:00 — Wrapping up

What to bring:
Just curiosity and questions. If you're working on applied AI, MLOps, model serving, or the cost and infrastructure behind it, we'd love to hear how you're approaching it.

Who this is for:
Data scientists, AI/ML engineers, data engineers, MLOps specialists, SREs, platform and infrastructure engineers, architects, and engineering leaders focused on running real-world AI efficiently.

Where to find us:
AWS Office: Mr. Treublaan 7, 1097 JS Amsterdam