MLcon London 2026
| title | author | length |
|---|---|---|
| Designing Production-Ready Agentic Systems: Lessons from the Field | Francesco Esposito | 52m |
Production agents run in voice pipelines with strict latency and cost controls: time-to-first-token improves after disabling Azure OpenAI sync filters, fillers… | ||
| An Agent Walks Into a Bar... and Meets Another Agent: Multi-Agent Systems in Action | Maxim Salnikov | 45m |
Argues that scaling from one LLM-driven chatbot to coordinated teams avoids prompt bloat and enables specialization, illustrating four concrete orchestration… | ||
| Stop Paying for Frontier Models | Rachel Nabors | ? |
Most of us reach for a frontier model by default and pay for it on every call, in latency, in energy, in cash, and in everything that leaves their stack. For… | ||
| LUNA 2 0 Multi Agent Edge AI with a Local Decision Engine | Philip Oonosy, Angelie Jane | 54m |
LUNA is a working Raspberry Pi + tablet prototype that runs a full edge multi‑agent stack, using the OpenAI Responses API, MCP, LangGraph configs and Azure for… | ||
| Structured Outputs in OpenAI: Building Reliable JSON-Based AI Integrations | Rainer Stropek, Melanie Bauer | 47m |
Shows production patterns that force LLMs to emit JSON via the OpenAI responses API using developer-defined JSON schema so outputs are type-safe and reduce… | ||
| Confidential AI: Enabling Trustworthy Ecosystems in the Age of Agentic Intelligence - Session | Yatindra Shashi | 37m |
Argument: use hardware TEEs to protect data-in-use, not just rest/transit; claim is that Intel SCX/TDX plus attestation and TDX Connect for GPUs let… | ||
| Attention is not all you need - Keynote | Christoph Henkelmann | 34m |
Argues that LLMs are powerful but not universal, showing concrete alternatives: TP PFM pretrains transformers for spreadsheets, T-Rex and XLSTM return strong… | ||
| Experience with an end-to-end MLOps pipeline: Practical insights and 3D architectural perspectives | René Brunner, Eric Joachim Liese | 53m |
A production MLOps stack evolves from glue code and modules rather than a monolith: MLflow + Aporia for model registry and monitoring, Kafka → Spark → S3… | ||
| nothing matches | ||