Skip to content

Details

Stop Babysitting LLMs, Start Shipping.

PRs got bigger, not better. Seniors are burning hours on AI bug hunts while throughput stays flat. Encode your existing SDLC into a gated multi-agent flow so LLMs do the noisy work. Review changes that have already passed CI.

## Swarm Architecture

Encodes a full SDLC loop (issuespec → plan → PRgatedeploy) in a gated 45-agent swarm you can explore after the talk.

## Shift Left

Mutation-on-diff and contract/schema checks wired into your agentic flow before review. Reduce review pain instead of just generating more code.

## Failure Modes & Guardrails

Hallucinations, reward hacking, process confabulation, and how your architecture catches and fixes them before they reach reviewers.

## Takeaways

  • ✓You'll leave with a working demo swarm repository you can clone into your own repos, and a clearer way to decide when to lean on tokens versus human review.
  • ✓For platform and DevEx engineers, tech leads and staff+ developers, and engineering leaders who are tired of wrestling with AI-generated code and review pain.
  • ✓The talk and swarm will be released under Apache 2.0.

Come at 5:30 pm have some pizza and chat for a bit. The presentation starts at 6:00 pm.

Related topics

Events in Toronto, ON
Artificial Intelligence
Microsoft Azure
Data Center and Operations Automation
DevOps
Github

You may also like