PCD October 2026

Welcome to another meetup brought to you by our community speakers always eager to share their work and findings on AI tooling.

Register and ignore the waiting list, since we have plenty of room. I'll accept your RSVP.

Please join us in person and let's have an amazing learning and networking event at our host "Neapolis University Paphos".

Speakers wanted:
If you are working on something you would like to share or you are playing / learning something new, let us know. Let's keep the community alive by sharing and learning from each other.

Location:
Neapolis University Paphos
2 Danais Avenue, Paphos 8042

ROOM: Amphitheater

Schedule:
18:50 Doors opening
19:00 Welcome Intro
19:10 Talk by Neo
19:50 Talk by Tom Klein
20:40 Networking
21:00 Closing

Talk#1 Details: "The AI Model Is Only Half the API"

How routing, observability and different inference paths shape the cost, speed and reliability of AI APIs.

When we call GPT, Claude, Gemini or another AI model, we usually think the model itself defines the API experience. In practice, the same model can be served through different inference paths — official APIs, managed capacity, third-party inference infrastructure and other sources — each with different cost, latency, throughput and reliability.
This creates an interesting engineering problem: if delivery quality changes over time, how do we know when a path is getting worse? When should traffic move somewhere else? And if there is no single “best” route for every workload, how should developers express what they actually care about?
In this talk, Neo will share lessons from building 1API, an AI API delivery platform. We’ll look at observability signals such as time to first token and throughput, health grading, routing decisions, fallback, session affinity, and the idea of Delivery Lanes — user-facing policies that balance cost, performance and stability across different inference paths.
The goal is not to introduce another OpenRouter. It is to explore what happens after the model has already been chosen, and why the delivery layer is becoming an important part of production AI infrastructure.

Neo has spent the past 20 years building software, with a focus on architecture, distributed systems, and high-concurrency infrastructure. Today, he is a heavy user of Codex and Claude Code, running roughly 15–40 billion tokens per month through frontier models. Working at that scale exposed the cost, latency, and reliability problems behind AI APIs, which led him to build 1API, an AI delivery and routing platform.

Talk #2 Details: "From Automation to SaaS in One Talk"

A tax return doesn't just cost money – above all, it costs time. That's exactly what Tom built TaxItEasy to fix. What began as a simple idea and an n8n automation running through a Telegram bot is today a real tax software for Europe. In his talk, Tom speaks openly about the challenges in the build process – test-driven development, AI hallucinations, and the constant evolution of the product as requirements from users and tax advisors keep shifting.

Tom Klein is the founder of TaxItEasy, a pan-European SaaS that uses AI to turn invoices and receipts into structured accounting data for freelancers and SMEs. Alongside his own product, he consults for one of the largest tax software providers on the market. His background spans n8n automation, LLM engineering, and Operations Research — he's a PhD candidate at TU Munich. Based in Paphos, Cyprus.