OSS + ELC EU 2026: Full Schedule

7-9 October, 2026
Prague, Czechia
View More Details & Registration
Important Note: Timing of sessions and room locations are subject to change.

The Sched app allows you to build your schedule but is not a substitute for your event registration. You must be registered for Open Source Summit Europe 2026 to participate in the sessions. If you have not registered but would like to join us, please go to the event registration page to purchase a registration.

11:20 CEST

Context Engineering: Connecting the Dots With Graphs - Stephen Chin, Neo4j

Wednesday October 7, 2026 11:20 - 12:00 CEST

Forum Hall (Floor 2)

AI systems need more than intelligence; they need context. Without it, even the most advanced models can misinterpret information, lose track of details, or arrive at conclusions that don’t hold up. Context engineering is emerging as a discipline that shapes how AI perceives, recalls, and reasons about information.

This talk will explore how context provides the foundation for reasoning, problem solving, and explainability in AI. We will look at techniques such as connected memory, contextual retrieval, and graph-based knowledge representation that give large language models a more reliable way to connect information and draw logical conclusions.

Attendees will come away with a practical understanding of how to design effective context pipelines that align AI with real-world knowledge and user intent, and why context engineering is becoming a central part of building trustworthy and impactful AI systems.

Speakers

Stephen Chin

VP of Developer Relations, Neo4j

Stephen Chin is VP of Developer Relations at Neo4j, conference chair of the LF AI & Data Foundation, and author of numerous titles including the upcoming GraphRAG: The Definitive Guide for O'Reilly. He has given keynotes and main stage talks at numerous conferences around the world... Read More →

Wednesday October 7, 2026 11:20 - 12:00 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Any

13:30 CEST

Off the Laptop, Into Production: An Open Source Stack for AI Agents on Kubernetes - Brian Hammons & Nirmal Mehta, Amazon Web Services

Wednesday October 7, 2026 13:30 - 14:10 CEST

Forum Hall (Floor 2)

AI agents are fast becoming how modern software gets built. But taking one from prototype to safe, production-grade operation is still genuinely challenging, and how they behave at scale remains largely uncharted. Agents are stochastic, so testing becomes continuous evaluation. They run code nobody wrote, so a shared kernel stops being a boundary you can trust. They decide their own network calls at runtime, so egress is no longer predictable. And they still need what any service needs: source control, CI/CD, observability, memory, and tool access.

This session walks the full lifecycle of running agents on Kubernetes, open source the whole way. The "outer loop" (build, deploy, trace, evaluate) runs on tools like GitLab, Argo, Langfuse, Milvus, and MCP gateways. The "inner loop" (isolated execution) runs on the kubernetes-sigs Agent Sandbox CRD, gVisor, and FQDN-aware egress via Cilium. We'll show what maps cleanly from the platform you run, what changes for agents, and the production gotchas no quickstart warns you about, drawn from building these in the open with teams running from a handful of developer sandboxes to the hundreds of thousands that frontier-scale RL spins up.

Speakers

Brian Hammons

Principal Solutions Architect, AWS

Brian Hammons is a Principal Architect at AWS, focused on Kubernetes-native AI agent development and platform engineering. He led the first AWS-backed contribution of the kubernetes-sigs Agent Sandbox CRD into the open source awslabs/ai-on-eks project. An Amazon EKS launch-team member... Read More →

Nirmal Mehta

World Wide Tech Leader - Containers, Amazon Web Services

Nirmal Mehta is the World Wide Tech Leader for Containers at AWS and a Principal Specialist SA in the Worldwide Application Modernization team. He is experienced in open source platform engineering, kubernetes, MLOps, agentic devops and executive org strategy. He has presented at... Read More →

Wednesday October 7, 2026 13:30 - 14:10 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Intermediate

14:25 CEST

Agent Gateway: The One Decision That Eliminates AI Engineering Complexity - Lin Sun, Solo.io

Wednesday October 7, 2026 14:25 - 15:05 CEST

Forum Hall (Floor 2)

As AI systems move from experimentation to production, engineering teams face a growing number of infrastructure decisions. How do you secure MCP servers without modifying them? How do you seamlessly transition and fail over across multiple LLM providers? How do you enforce token rate limits and usage policies for LLMs? How do you govern agent-to-agent and agent-to-MCP communication, control context growth, enforce security policies, observe traffic, and scale operations across environments?

Rather than solving each challenge independently, organizations can adopt a single architectural pattern: the agent gateway.

Through live demos, Lin introduces the agent gateway pattern and demonstrates how it simplifies AI infrastructure by eliminating the need for teams to repeatedly solve the same cross-cutting concerns. Using agentgateway, an open-source implementation of the pattern, she will show how to secure and federate MCP servers without modifying them, route, fail over, and rate limit LLM traffic across providers, enable advanced capabilities such as code mode and progressive disclosure, and operationalize AI workloads on Kubernetes.

Speakers

Lin Sun

Head of Open Source, Solo.io

Lin is the Head of Open Source at Solo.io, co-chair of KubeCon + CloudNatibeCon 2026. She actively contributes to AAIF and CNCF projects. She is the author of “Sidecar-less Istio Explained” and “AI Agents in Kubernetes”, and holds more than 200 patents.

Wednesday October 7, 2026 14:25 - 15:05 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Any

15:35 CEST

Iceberg for Agents - Turning Lakehouse Data Into AI-Ready Context - Andrew Madson, Fivetran

Wednesday October 7, 2026 15:35 - 16:15 CEST

Forum Hall (Floor 2)

AI agents fail in production because they're overwhelmed with data but starved for context. LLM models aren’t the problem. The bottleneck is the data stack: fragmented silos, inconsistent definitions, and logic hidden in tribal knowledge. Agents need structured, reliable, and interpretable context—not just data access.

In this session, we'll show how Apache Iceberg becomes the backbone of AI-ready pipelines. You’ll learn how to elevate your Iceberg implementation from a storage format to a live context layer that powers structured retrieval-augmented generation (RAG), schema-aware agents, and autonomous reasoning grounded in truth.

What we’ll cover:
1. Iceberg Foundations for AI - from ACID to Time Travel
2. From Rows to Relationships - The role of the semantic layer
3. Structured RAG in Practice - Fully open source

The session includes a live demo of a fully open-source Structured RAG stack built on Apache Iceberg, featuring semantic query translation, hybrid retrieval, and governed agent reasoning. Expect architecture diagrams, real code, and practical guidance.

Speakers

Andrew Madson

Head of Developer Relations, Fivetran

Andrew Madson is Head of Developer Relations at Fivetran, where he builds global programs that help developers and data teams adopt modern data and AI tooling. He previously led DevRel and evangelism at Tobiko Data (creators of SQLMesh) and Dremio, is author of the O'Reilly Definitive... Read More →

Wednesday October 7, 2026 15:35 - 16:15 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Beginner

16:30 CEST

From Single Agent To Production Swarm: Patterns for Building Reliable Multi-Agent Systems - Betty Zheng, AWS & Trista Pan, Tetrate

Wednesday October 7, 2026 16:30 - 17:10 CEST

Forum Hall (Floor 2)

AI agents are evolving from simple chatbots to coordinated systems that plan, collaborate, and execute real-world tasks. But moving from a working prototype to a production-grade multi-agent system is where most teams struggle.

In this session, two practitioners share battle-tested patterns from deploying agent swarms in industrial environments. We cover:
The architectural shift from single-agent to multi-agent: when and why you need a swarm
Comparing open source frameworks (LangChain, AutoGen, Strands Agents SDK) — strengths, trade-offs, and selection criteria
Practical orchestration patterns: task delegation, shared memory, tool coordination via MCP (Model Context Protocol)
Live demo: building a multi-agent system that collaborates to solve a real task
Safety-first design: guardrails, execution boundaries, and observability for production
Attendees will leave with a reusable architectural blueprint applicable to any open source agent framework.

Speakers

zheng yubin

Senior developer advocate, AWS

Yubin Zheng is a seasoned engineer with over 20 years of experience in ICT and digital transformation. She focuses on building real-world AI systems, including agent runtime design, multi-agent coordination, and production deployment. She is an active contributor to developer and... Read More →

Trista Pan

AI Engineer, Tetrate

AI Engineer at Tetrate, Ex AI Product Director at Bytedance, Former Co-founder of SphereEx, Apache Member and Incubator Mentor, and AWS Data Hero

Her expertise is in big data and AI. She was elected an OSCAR Top Open Source Pioneer. In 2022, her paper “Apache ShardingSphere: A Holistic and Pluggable Platform for Data Sharding” was published on ICDE. As the first author, she wrote a book titled A Definitive Guide to Apache S... Read More →

Wednesday October 7, 2026 16:30 - 17:10 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Advanced

17:25 CEST

Nine Seconds To Production: Security Governance for AI Agents - Jasdeep Singh Bhalla, GoDaddy

Wednesday October 7, 2026 17:25 - 18:05 CEST

Forum Hall (Floor 2)

On April 24, 2026, an AI coding agent deleted a company's entire production database in nine seconds. Every customer record, every reservation, and every backup vanished. The system prompt explicitly told the agent never to execute destructive commands. It did it anyway.

It was a chain of architectural failures: an agent improvising instead of stopping, overprivileged credentials, no human approval for high-risk actions, and backups inside the same blast radius.

This session recreates the incident and rebuilds the security architecture live. You'll see how OPA enforces policy before tools execute, Falco detects suspicious runtime behavior, and OpenTelemetry captures structured reasoning traces for compliance-grade audit logs.

We'll implement six defensive layers: pre-execution policy enforcement, runtime behavioral monitoring, structured audit logging, least-privilege execution roles, infrastructure deletion protection, and recovery systems isolated outside the agent trust boundary.

Prompts are guidance. Infrastructure is enforcement. You'll leave with a vendor-neutral security architecture for deploying AI agents safely across any framework or LLM provider.

Speakers

Jasdeep Singh Bhalla

Senior Software Engineer, GoDaddy

Jasdeep Singh Bhalla is a Senior Software Engineer at GoDaddy, specializing in AI security governance, cloud native infrastructure, and distributed systems. With over a decade of experience across GoDaddy, Electronic Arts, Dialpad, and Yahoo, he has architected production platforms... Read More →

Wednesday October 7, 2026 17:25 - 18:05 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Intermediate

10:50 CEST

Just Keep Swimming: Benchmarking OSS Agent Memory on Kubernetes With OpenTelemetry - Henrik Rexed, Dynatrace

Thursday October 8, 2026 10:50 - 11:30 CEST

Forum Hall (Floor 2)

Your AI agent is Dory. Every conversation it wakes up unable to remember a thing , so you, Marlin, re-explain everything every turn. Four open-source memory systems promise to fix that: Cognee, MemOS, Honcho, and MemPalace. But which architecture actually pays off in tokens, dollars, and resolved tasks?
This talk runs two experiments. First, we benchmark all four backends head-to-head on a single Kubernetes substrate , same harness, same workloads, same OpenTelemetry Collector and rank them by token efficiency, cost per resolved task, and recall under contradiction.
Then we test the question most teams haven't asked: is running an agent with memory worth it at all? We measure two real OSS agent frameworks , kagent (CNCF Sandbox) and sympozium with their built-in memory turned on, then turned off. Real workloads, real numbers.
None of the four backends ships first-party OTel today; the Helm chart, OTel pipeline, and upstream PRs are how we close that gap for the ecosystem.

Speakers

Henrik Rexed

Cloud Native Advocate, Dynatrace

Henrik is a Cloud Native Advocate at Dynatrace, the leading Observability platform. Prior to Dynatrace, Henrik has worked more than 15 years, as Performance Engineer. Henrik Rexed Is Also one of the Organizer of the conferences named WOPR, KCD Austria and the owner of the Youtube... Read More →

Thursday October 8, 2026 10:50 - 11:30 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Intermediate

11:40 CEST

Building Real-Time, Data-Aware Intelligence With Postgres & Model Context Protocol (MCP) - Yogesh Jain, EnterpriseDB

Thursday October 8, 2026 11:40 - 12:20 CEST

Forum Hall (Floor 2)

Imagine asking an AI agent to analyze any data, only for it to hallucinate a schema that doesn't exist. This "context gap" is the primary barrier to reliable Data-Aware AI. While LLMs are brilliant reasoners, they are historically "blind" to the live, structured world of Postgres - until now!

In this talk, we will explore how the Model Context Protocol (MCP) bridges this gap, transforming Postgres from a passive storage layer into an active reasoning engine. We’ll dive into the architecture of a Postgres MCP server, demonstrating how it provides agents with secure, structured access to schemas and metadata. By utilizing tools like pg_catalog and EXPLAIN, we enable LLMs to understand table relationships and optimize performance before running queries.

Expect live demos and practical insights on making Postgres "talk" to AI safely and effectively.

Key Takeaways:

* How Postgres + MCP provides LLMs with correct, real-time database context.
* Leveraging Postgres metadata and schema to eliminate SQL hallucinationcs.
* Implementing secure, multi-tenant, and read-only access for AI agents.
* Enabling AI agents to use EXPLAIN and other tools for self-optimization.

Speakers

Yogesh Jain

Staff SDE at EDB, EnterpriseDB

Yogesh Jain is a Staff SDE at EnterpriseDB and a passionately curious full-stack developer. He specializes in building scalable solutions leveraging open-source technologies, with expertise spanning Kubernetes, observability, distributed systems, and cloud-native PostgreSQL.He is... Read More →

Thursday October 8, 2026 11:40 - 12:20 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Beginner

13:50 CEST

Yukti: A Unified Inference Interface for Low-Latency Machine Learning in High-Energy Physics - Sanjiban Sengupta, CERN, University of Manchester

Thursday October 8, 2026 13:50 - 14:30 CEST

Forum Hall (Floor 2)

Machine learning is increasingly used in high-energy physics, particularly in trigger systems that process data at rates of 100 kHz while making real-time event selection decisions. The latency and reliability requirements demand highly optimized inference pipelines. While portable solutions such as ONNX Runtime simplify deployment, many applications rely on hardware-specific libraries like NVIDIA TensorRT and MIGraphX (via ROCm) for optimal performance. Code-generation approaches such as SOFIE offer additional efficiency but introduce integration complexity.

We present a unified inference interface that abstracts backend-specific details while preserving performance. It enables execution across multiple inference libraries without data copies or user-side configuration changes. An offline processor converts trained models into backend-optimized plans, and a lightweight runtime loads and executes them through a common API with direct data access for heterogeneous environments.

Speakers

Sanjiban Sengupta

Doctoral Student, CERN, University of Manchester

Sanjiban is a Doctoral Student at CERN, affiliated with the University of Manchester, researching ML inference optimization for the LHC. He contributed to SOFIE, focusing on Keras/PyTorch parsing, ONNX-based operators, and GNN support. He was a CERN Summer Student (2022) and a GSoC... Read More →

Thursday October 8, 2026 13:50 - 14:30 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Intermediate

14:40 CEST

Designing Permissioned AI Agents That Can Run Offline - Miley Fu, Olares Tech

Thursday October 8, 2026 14:40 - 15:20 CEST

Forum Hall (Floor 2)

AI agents are moving from chat interfaces into workflows that read files, call tools, modify state, and coordinate multiple services. In cloud-hosted designs, this often creates a fragile trust boundary: private data leaves the device, tools are authorized through broad API keys, and the user has limited visibility into what the agent can do. This talk presents an open-source, local-first architecture for permissioned AI agents that can operate offline on Linux edge or personal infrastructure. We will break down the architecture into model serving, retrieval, tool invocation, capability scoping, identity, audit logs, sandboxing, human approval, and fallback synchronization when connectivity returns. The session is not about a specific agent framework; it is about the system boundary around agents. Attendees will learn how to separate reasoning from action, how to map tools to least-privilege permissions, how to keep sensitive context local, and how to evaluate agent workflows when network access is unavailable or intentionally disabled.

Speakers

Miley Fu

Developer Relations Director, Olares Tech

Miley is the co-chair and keynote speaker for KubeCon+Open Source Summit and AI Dev China 2024. She works on WasmEdge runtime under Linux Foundation as the founding member for over 6 years. She talks at KubeCon, KCD Beijing+Shenzhen, CloudDay Italy, DevRelCon, Open Source Summit Japan... Read More →

Thursday October 8, 2026 14:40 - 15:20 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Any

15:50 CEST

Open Standards for Measuring Carbon, Energy, and Water in Software and AI - Navveen Balani, Linux Foundation

Thursday October 8, 2026 15:50 - 16:30 CEST

Forum Hall (Floor 2)

Software’s environmental footprint, especially AI’s, is becoming a business, governance, and regulatory priority. The EU AI Act introduces environmental transparency expectations for general-purpose AI, while the CSRD expands sustainability reporting obligations to 50,000+ companies. As software and AI adoption accelerate, organizations are increasingly seeking consistent ways to measure carbon, energy, and water impacts across workloads and inference.

The Green Software Foundation, a Linux Foundation project, is helping shape the open standards foundation for this transition:

• SCI (ISO/IEC 21031:2024): the ISO-ratified standard for software carbon intensity
• SEE: measuring software energy efficiency across workloads
• SCI for AI: extending measurement to AI training, fine-tuning, and inference
• SWE: addressing the growing water footprint of software and AI systems

Drawing on direct experience chairing and co-leading these specifications, this session explores how these standards connect, align with emerging expectations such as the EU AI Act and CSRD, and help organizations build measurable, transparent, and sustainable software and AI systems.

Speakers

Navveen Balani

Executive Director , Green Software Foundation, Linux Foundation

Navveen Balani is the Executive Director of the Green Software Foundation and a technology leader with 25+ years of experience building exponential and responsible technology systems. He chaired the creation of the ISO-ratified Software Carbon Intensity (SCI) standard and leads SCI... Read More →

Thursday October 8, 2026 15:50 - 16:30 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Beginner

16:40 CEST

Panel: The Future of AI Inference in Cloud Native: Open Source Innovation, Challenges, and Opportunities - Martin Hickey, IBM Research; Maroon Ayoub, Red Hat; Nili Guy, IBM; Cong Liu, Google; Hyunkyun Moon, Moreh

Thursday October 8, 2026 16:40 - 17:20 CEST

Forum Hall (Floor 2)

As AI workloads move into production in cloud native environments, the open source community must define the future of AI inference infrastructure. This panel convenes experts from leading projects to assess the ecosystem and chart a path forward.

Key themes include:

- The evolution of cloud native inference (from serverless to large‑scale distributed serving)
- Performance and efficiency advances such as KV cache management, quantization and hardware acceleration
- The need for standardization across APIs, model formats and interoperability.

They’ll also examine how projects like vLLM, llm‑d, and KServe can collaborate to reduce fragmentation.

Panelists will address real‑world challenges in Kubernetes deployment, including multi‑tenancy, GPU sharing, autoscaling and observability. There will also explore emerging trends such as edge inference, multi‑modal models and the convergence of training and inference systems.

Speakers

Martin Hickey

Senior Technical Staff Member, IBM Research

Martin Hickey is a STSM at IBM Research, focused on Open Source, Cloud Native Computing, and AI. Martin has notable contributions to open source projects like vLLM, LMCache, Kubernetes, Helm, OpenTelemetry and OpenStack. Martin is a core maintainer for LMCache and an emeritus core... Read More →

Maroon Ayoub

Senior Principal Machine Learning Engineer, Red Hat

Maroon Ayoub is an AI infrastructure architect at Red Hat focused on distributed inference. He co-leads development of llm-d and specializes in scaling LLM inference with Kubernetes-native architectures, KV-cache orchestration, and open source integrations.

Nili Guy

R&D and Senior Technical Staff Member, IBM

Nili is a Research Manager and Senior Technical Staff Member at IBM Research, co-creator of llm-d, and an expert in distributed inference and Kubernetes-native AI systems. She has led key open-source and productized inference initiatives across IBM’s AI platforms.

Cong Liu

Software Engineer, Google

Cong is a software engineering working on Google Kubernetes Engine (GKE), building solutions for AI Inference workloads, such as efficient request scheduling for Large Language Models (LLMs). His past experience include GKE release channels and safe cluster upgrades.

hyunkyun moon

ML Platform Engineer, Moreh

Hyunkyun Moon is an ML Platform Engineer at Moreh, where he builds high-performance LLM inference platforms using llm-d. He is an active contributor to open-source projects, including llm-d and vLLM. Previously at LINE Plus, he specialized in developing large-scale DBaaS using Kubernetes... Read More →

Thursday October 8, 2026 16:40 - 17:20 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Any

11:05 CEST

Scaling OSS Code Quality With Multi-Agent Swarms - Vikram Vaswani, Independent

Friday October 9, 2026 11:05 - 11:45 CEST

Forum Hall (Floor 2)

Development teams are seeing a rapid increase in code contributions, many of them from AI tools and pair-programming copilots. As AI-generated code increases, human reviewers can quickly get overwhelmed.

The instinct to throw "an AI reviewer" at the problem mostly doesn't work. A single agent produces shallow comments, misses context, and either over-flags or under-flags depending on how it's prompted.

This talk presents a pattern that works better, built and demonstrated from scratch: three specialized reviewing agents in parallel, followed by a developer agent that applies the fixes. One reviewer focuses on architecture, one on security, one on test coverage. When all three finish, a fourth agent rewrites the code to address each issue and opens a new PR for human sign-off.

The architecture is adaptable: teams with data sovereignty requirements can swap the hosted model for a self-hosted open-weight alternative (Llama 3, Mistral) served via vLLM or Ollama.

Attendees leave with a concrete, working pattern they can adapt to their own projects, and a clearer picture of where AI can genuinely reduce review bottlenecks.

Includes live demo.

Speakers

Vikram Vaswani

Developer Advocate

Friday October 9, 2026 11:05 - 11:45 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Any

11:55 CEST

Look MA, No GPUs! Experiment and Develop LLMs Without GPU Support - Alexon Oliveira, Red Hat

Friday October 9, 2026 11:55 - 12:35 CEST

Forum Hall (Floor 2)

It's hard to experiment, let alone develop applications, for LLMs when access to GPUs is restricted. But what if you could do your development on CPUs? This session demonstrates how to deploy vLLM on a local Kubernetes cluster using Kind and Podman Desktop, running entirely on a CPU-only laptop.

The presentation covers building a minimal environment, deploying vLLM containers, and running inference against models like Gemma-3 and Qwen2.5. A live benchmark comparison between vLLM and llama.cpp reveals when each runtime excels across different model sizes and hardware profiles. The session then extends the workflow to production-grade serving with KServe, including autoscaling and model lifecycle management.

Attendees will learn how to deploy and test vLLM locally without GPUs or cloud resources, understand when to choose vLLM versus llama.cpp based on model size and hardware, and extend a local setup to production-grade serving with KServe.

Speakers

Alexon Oliveira

Senior Principal Technical Account Manager, Red Hat

Alexon works as a Senior Principal Technical Account Manager at Red Hat focusing on Infrastructure and Management, Integration and Automation, Cloud Computing, Storage, and AI Solutions. He also contributes to produce and enhance documentation, knowledge-base articles, blog posts... Read More →

Look MA, No GPUs Experiment and Develop LLMs Without GPU Support pdf

Friday October 9, 2026 11:55 - 12:35 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Intermediate

14:00 CEST

Lightning Talk: 97% of MCP Tools Have Quality Defects & Here's How We Measured Them - Om Shree, Shreesozo

Friday October 9, 2026 14:00 - 14:10 CEST

Forum Hall (Floor 2)

Silent failures in MCP tool invocation aren't random. They trace back to a single root cause: tool descriptions that are too vague for agents to reliably select the right tool.
A SAIL Research study of 856 tools across 103 MCP servers found 97% have at least one quality defect, 56% fail to clearly state what the tool does, and 89% provide no guidance on when not to use it. A second study of 10,831 servers confirmed that well-written descriptions get selected 260% more often, improving task success rates by ~6 points.
Working with the Glama founder, I developed TDQS (Tool Definition Quality Score) : an open-source framework scoring MCP tools across six dimensions: Purpose Clarity, Usage Guidelines, Behavioral Transparency, Parameter Semantics, Conciseness, and Contextual Completeness.
This talk covers how TDQS was built, what scoring thousands of real-world servers revealed, and how the open-source community can adopt it as a quality standard for MCP tool authorship.

Speakers

Om Shree

Founder & MCP Infrastructure Researcher, Shreesozo

I'm the founder of Shreesozo, an AI content studio focused on MCP and agentic AI. I've written 100+ technical pieces for Glama.ai and Gentoro, covering everything from protocol internals to real-world agent deployments. I run MCP Weekly, published on YouTube (1.2K subscribers) and... Read More →

Friday October 9, 2026 14:00 - 14:10 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Intermediate

14:10 CEST

Lightning Talk: Real Observability for Self-Hosted LLMs on Ubuntu - Shardul Deshpande, Canonical

Friday October 9, 2026 14:10 - 14:20 CEST

Forum Hall (Floor 2)

Running LLMs near devices or in constrained edge environments changes the observability problem. A small model can answer successfully while still being unusable: first token arrives too late, prompts burn scarce CPU/RAM budget, output is malformed, or gateway/model-selection errors hide behind green Linux and container metrics.

This session walks through a reproducible, CPU-first Ubuntu demo for resource-constrained self-hosted LLMs. It uses small Ollama models (llama3.2:1b and qwen3.5:2b) on the same CPU workload with MicroK8s, Juju, COS Lite, Tempo, a FastAPI OpenAI-compatible gateway, and Canonical's Charmed OpenTelemetry Collector. The gateway emits OpenTelemetry GenAI spans, metrics, and trace-correlated logs; Prometheus, Loki, Tempo, Grafana, and Alertmanager turn them into an LLM golden-signals view.

Using recorded normal, slow, expensive, error, and low-quality scenarios, attendees will see how time-to-first-token, token usage, traces, alerts, and quality signals reveal whether a local model is useful within tight compute constraints. The model comparison shows why embedded and edge AI decisions need measured latency/cost/quality trade-offs, not just 'the model runs.'

Speakers

Shardul Deshpande

Software Engineer, Observability Stack, Canonical

Shardul Deshpande works at Canonical on the Observability Stack (COS), building practical demos around observability, Kubernetes, and self-hosted AI infrastructure. His work connects Ubuntu, MicroK8s, Juju, OpenTelemetry, and Grafana-based operations so teams can measure whether AI... Read More →

Friday October 9, 2026 14:10 - 14:20 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Intermediate

14:20 CEST

7 MCP Servers, 50+ Tools, Zero Duplicate Code: A 5-Minute Architecture Tour - Harika Chebrolu, IBM

Friday October 9, 2026 14:20 - 14:30 CEST

Forum Hall (Floor 2)

We had a problem: every new integration meant writing the same code twice—once for our React dashboard, once for our AI assistant. JIRA, TestRail, Jenkins, GitHub—the duplication was unsustainable.

MCP changed everything. In this lightning talk, I'll show how we built 7 MCP servers that now power both our dashboard AND our AI agents from a single source of truth. 50+ tools, used daily by Dev, QE, and Management teams.

In 5 minutes, you'll see:

- The architecture: FastAPI as a unified backend → MCP Servers → External APIs (JIRA, TestRail, Jenkins, GitHub, GSheets, Release Calendar, Risk Predictor)
- The pattern: Thin MCP tools that call centralized APIs—one integration, two consumers
- Real metrics: 10+ hours/week saved per engineer; release tracking, regression monitoring, and production health checks unified.
- Why MCP wins: AI agents and dashboards evolve independently while sharing the same data layer.

We eliminated duplicate code, reduced maintenance burden, and shipped faster. All code is open source.

Speakers

Harika Chebrolu

Software Engineer | Multi-Agent Systems, IBM

Harika Chebrolu is a software engineer building autonomous systems for infrastructure operations. She architected a multi-agent platform managing 200+ storage nodes and extracted MCP servers for safe, interoperable infrastructure tooling. Her open-source work includes multi-agent... Read More →

Friday October 9, 2026 14:20 - 14:30 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Intermediate

14:50 CEST

Building Fine-Grained Permissions for AI Agents - Sohan Maheshwar, AuthZed

Friday October 9, 2026 14:50 - 15:30 CEST

Forum Hall (Floor 2)

This talk could prevent the next big data breach.

Building enterprise-ready AI poses challenges around data security, scalability, and integration, especially in compliance-regulated industries. We're already seeing security breaches with AI Agents in the news. This is a complex problem - Imagine having N users, M Agents and O actions. How do you design permissions around that?

This session will cover how modern permissions systems can ensure AI Agents have access only to authorized data. The talk will look at why the Google Zanzibar model of authorization which uses Relationship-Based Access Control (ReBAC) is well suited for fine-grained authorization at scale. The talk covers the nuts and bolts of how a Google Zanzibar system works under the hood, and how to apply it to AI Agents with techniques such as pre-filteration and post-filteration.

The talk will also include a live code demo implementing authorization for AI Agents + RAG using Open Source tools such as Weaviate, Langchain, and SpiceDB.

Speakers

Sohan Maheshwar

Lead Developer Advocate, AuthZed

Sohan is a Lead Developer Advocate at AuthZed, based in the Netherlands. He started his career as a developer building mobile apps and has been living in the cloud since 2013, in companies such as Amazon, Fermyon and Gupshup. He is also an AAIF Ambassador and an O' Reilly author... Read More →

Friday October 9, 2026 14:50 - 15:30 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Advanced

15:40 CEST

When Coding Agents Cheat: Measuring the Security of AI-Generated Code With an Open Benchmark - Luca Compagna, Endor Labs & Danqing Wang, Carnegie Mellon University

Friday October 9, 2026 15:40 - 16:20 CEST

Forum Hall (Floor 2)

Vibe coding is now the norm, and more open-source code is AI-generated. But is it secure?

We present the Agent Security League, an open leaderboard built on SusVibes (200 real-world tasks, 108 OSS projects, 77 CWEs). Unlike benchmarks frontier labs cite to showcase progress, SusVibes evaluates what developers actually use: a harness coupled with an LLM, not the model alone. Across 15+ combinations — GPT-5.5, Gemini 3.5, Fable 5 — with agents like Cursor and Claude Code, functional correctness can exceed 80%, yet our best fair security score is 29%.

This means ~7/10 working AI patches leave the bug open.

Strikingly, even models (e.g., Fable 5) that lead model-only benchmarks score no better on security once wrapped in a real agent.

Agents also cheat: early on they reverse-engineered fixes from git history, inflating scores up to 42×. Our 3-layer pipeline—prompt hardening, workspace sanitization, LLM-based anti-cheating eval—nearly killed that, but memorization of upstream fixes from training data persists.

We show cheating examples, share fair results, what actually improves security, and argue anti-cheating and contamination controls must be standard for agentic benchmarks.

Speakers

Luca Compagna

Security Researcher, Endor Labs

Security Research Consultant at Endor Labs, after more than 15 years at SAP, researching and innovating in the areas of security testing, security engineering, and AI.

Regular presenter at industrial and community venues (OWASP) and leading security conferences, he has published... Read More →

Danqing Wang

PhD, Carnegie Mellon University

Danqing is a Ph.D. candidate at LTI in Carnegie Mellon University (CMU), advised by Prof. Lei Li. Her research focuses on Strategic Planning and Reasoning in LLM Agents, including agent collaboration, agent competition, agent communication and agentic safety.

Friday October 9, 2026 15:40 - 16:20 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Intermediate

16:30 CEST

From Hours To Minutes: Building Production-Ready Agentic Systems on Kubernetes - Purnanand Kumar & Dipali Chatterjee, IBM

Friday October 9, 2026 16:30 - 17:10 CEST

Forum Hall (Floor 2)

Agentic AI systems are rapidly evolving from research experiments into production workloads. However, while building a proof-of-concept agent can take only a few hours, building a reliable, scalable, and observable agentic platform remains a significant challenge.

This session explores how modern open source technologies can be combined to build production-ready agentic systems on Kubernetes. We will walk through a complete architecture that includes large language models, agent orchestration, retrieval systems, tool execution, memory management, observability, and infrastructure automation.

Using technologies such as Kubernetes, Kubeflow, LangGraph, Model Context Protocol (MCP), vector databases, and open-source serving frameworks, we will demonstrate how to move from a simple agent prototype to a resilient platform capable of supporting enterprise-scale workloads.

Speakers

Dipali Chatterjee

IBM

Purnanand Kumar

Advisory software engineer, IBM

Led the development efforts in creating robust and scalable back-end systems for various projects, focusing on storage domain solutions. Leveraged Golang to build high-performance applications, optimizing data processing and storage operations. Collaborated with cross-functional teams... Read More →

Friday October 9, 2026 16:30 - 17:10 CEST
Forum Hall (Floor 2)

Open AI & Data

Audience Experience Level Intermediate

11:20 CEST

13:30 CEST

14:25 CEST

15:35 CEST

16:30 CEST

17:25 CEST

10:50 CEST

11:40 CEST

13:50 CEST

14:40 CEST

15:50 CEST

16:40 CEST

11:05 CEST

11:55 CEST

14:00 CEST

14:10 CEST

14:20 CEST

14:50 CEST

15:40 CEST

16:30 CEST

Get help with the event