Overview
Conf42: MLOps 2026 is a free online conference for engineers and practitioners working with machine learning operations. The event takes place on Thursday, November 19, 2026, and all talks premiere on YouTube on the event day at 5:00 PM GMT. Recordings stay available on demand, so attendees can watch them at their own pace after the live premiere.
Who should attend
The conference is aimed at engineers and practitioners who build, deploy, and operate ML and LLM systems. Its program spans agent engineering, data pipelines, evaluation, governance, inference, and platform design.
Format
- Fully online and free to attend.
- Talks premiere live on YouTube on November 19, 2026, starting at 5:00 PM GMT.
- All talks remain available on demand afterward.
- Attendees can add the event to their calendar through Luma, Google Calendar, iCal, or Outlook, and can subscribe for updates.
Topics and tracks
The program is organized into several tracks:
- Agents: Multi-agent reliability, running large agent fleets in production, agent failure recovery, LangGraph.js operations, context and memory management, and orchestration.
- Data: Large-scale LLM data processing, data quality for ML at healthcare scale, data architectures for physical AI, and database DevOps for ML in finance.
- Evals: Testing LLMs that generate clinical configuration, eval-as-code, reproducible regression tests from agent traces, and treating evals as deployment artifacts.
- Governance: Tamper-evident audit trails for agents, compliance automation in MLOps, governance rules engines, and fairness in compliant ML models.
- Guardrails: Enforcing agent rules in code, graph-based retrieval to reduce hallucinations, and write-gates for agent fleets.
- Inference: Running vLLM on Kubernetes across clouds, edge inference with ExecuTorch, and GPU workload attribution.
- Models: Deep learning pipelines for medical imaging, anomaly detection for retail operations, and municipal bond intelligence.
- Platforms: Shared control planes for generative AI and ML platforms, self-healing cloud infrastructure for agentic AI, and microservices observability with agentic LLMs.
- Production: Rollback testing, enterprise agent deployment, and shipping LLM features at e-commerce scale.
- Security: Securing the AI stack, hardening LLM inference containers, identity and authorization for AI agents, and authorization evals in delivery pipelines.
Speakers
The lineup includes speakers from companies such as Amazon, AWS, Adobe, Cisco Systems, Fidelity, JPMorgan Chase, Microsoft AI, OpenAI, OWASP, PayPal, Roblox, T-Mobile, Uber, UK Ministry of Defence, Walmart Global Tech, and Yelp, along with independent practitioners and several stealth-mode teams. Example speakers and talks include Daniel Akhabue on agent protocol engineering, Muhammad Haris Farooq on running 50+ AI agents in production, Reza Amini on agent failure recovery, Phoebe Wang on ML workflows with Codex, Flyte, and MCP, Ocean Coldiron on evaluating an LLM that writes clinical configuration, and Joao Marques on tamper-evident audit trails for AI agents.
Sponsors and community
The event is supported by diamond and gold sponsors and media partners. Attendees can join the Conf42 community for event notifications, a newsletter, and discount codes for in-person events. Conf42 also runs a Discord community and publishes its terms, code of conduct, and privacy policy on its website.
Note on the schedule
The conference page indicates that more talks will be announced soon, so the final program may expand beyond the talks currently listed.




