Manychat SRE Meetup

Dark Side of AI Infrastructure - What It Takes to Keep Systems Running

📍 Cinema The Pulse, Amsterdam Zuid📅 November 19, 2026🕕 18:00–22:00 (Doors open at 18:00, Talks start at 18:30)

💡 About

AI gets the headlines. Infrastructure does the heavy lifting.

Behind every new model, product and platform are engineering teams making difficult decisions about performance, reliability, capacity and scale.

But what does it actually take to keep complex systems running?

Manychat is bringing together Amsterdam's SRE, Platform Engineering and Infrastructure community for an evening of technical talks with engineers from Manychat, JetBrains and Nebius.

We'll get into the challenges that rarely make it into product announcements: responding to incidents, making the most of limited GPU capacity, managing infrastructure trade-offs and keeping production systems reliable.

Three engineering perspectives. Real production challenges. Plenty to discuss.

And once the talks are over, we'll continue the conversation over food and drinks.

⚡ What to Expect

🎤 3 Technical Deep DivesEngineering insights from the teams behind Manychat, JetBrains and Nebius.

⚙️ Behind the InfrastructureIncident response, GPU scheduling, capacity constraints and the decisions that shape production reliability.

💬 Questions Worth AskingGo beyond the slides, discuss technical trade-offs and compare approaches with speakers and fellow engineers.

🥂 Meet the CommunityConnect with Amsterdam's SRE, Platform and Infrastructure engineers over drinks and snacks.

🗓️ Agenda

18:00 – Doors open, food & networking

18:30 – Welcome & opening

18:40 – 🎤 Dzmitry Yatskevich, Head of Infrastructure at Manychat

Cut Incident Time from 10 Hours to 10 Minutes

Twenty years of incident management distilled into practical lessons. Dzmitry will cover the stages of incident response, the OODA loop and approaches to reducing MTTA and MTTR - with or without additional tooling.

19:15 – 🎤 Andrei Vishniakov, MLOps Lead at JetBrains

Overcommit on Purpose: Fast Pulls, Dense Packing, and Fair GPU Scheduling for ML Workloads

Andrei will explore how to balance GPU availability and utilization, including the scheduling challenges of reinforcement-learning workloads and the engineering trade-offs involved.

19:45 – 🎤 Ivan Bakurskii, Lead SRE at Nebius

25+ PiB and Counting: Lessons from Running Nebius’s Data Platform.Ivan will share how Nebius runs a 25+ PiB internal data platform powered by YTsaurus, exploring unexpected challenges across Kubernetes and Linux, engineering trade-offs, and practical lessons from keeping distributed storage and data processing reliable in production.

20:20 – Closing

20:30 – 🥂 Drinks, snacks & community networking

22:00 – Event close

🎯 Built for...

Site Reliability Engineers

Platform Engineers

Infrastructure & Cloud Engineers

DevOps Engineers

Staff & Principal Engineers

Engineering Managers & Technical Leaders

MLOps & AI Infrastructure Engineers

For the engineers behind the systems - the ones building, scaling, monitoring and keeping them running.

🎟️ Registration

Attendance is free, but seats are limited.

Bring your questions, meet the people solving similar challenges, and leave with ideas worth taking back to your team.