Resilient Agents - Online Hackathon
Infrastructure fails. Rate limits hit. Timeouts happen. Most agents just crash.
Resilient Agents is a one-week online hackathon hosted by TrueFoundry with support from AWS Bedrock. The challenge: build an agent that keeps working when providers slow down, rate limits hit, tools fail, or model calls break.
Timeline
Kickoff: June 1, 2026 at 10:00 AM Pacific Time
Submissions close: June 7, 2026 at 11:59 PM Pacific Time
All hackathon communication will happen in Discord.
Discord: https://discord.gg/7dHQAsQq66
Channel: june-1-2026-resilient-agents-online-hackathon
The submission website will be posted shortly in the Discord channel.
The Challenge
Use TrueFoundry AI Gateway to connect to AWS Bedrock models, route LLM calls, and configure fallbacks. If your agent uses tools, add MCP Gateway and Guardrails so tool access is controlled and risky actions can be checked before they run.
Then stress-test it against failures like:
rate limits
model or provider outages
slow responses
tool failures
bad intermediate outputs
cascading errors across multiple steps
What You'll Build With
AWS Bedrock: access foundation models from providers like Amazon, Anthropic, Meta, Mistral AI, Cohere, and others through a managed AWS service.
TrueFoundry AI Gateway: route requests, configure fallbacks, monitor usage, and keep LLM calls governed from one place.
TrueFoundry MCP Gateway: give your agent safe, centralized access to tools and data sources. Use it to expose only the tools your agent needs, manage auth centrally, and keep an audit trail of tool usage.
Guardrails: add safety checks around LLM calls and MCP tool calls. Use them to block unsafe inputs, redact sensitive data, validate tool arguments before execution, and inspect tool results before the model sees them.
What Judges Care About
AI Gateway setup: routing, fallback, observability, and control
MCP Gateway usage: safe tool access, scoped permissions, auth, and auditability
Guardrails: checks that block, redact, or validate risky LLM and tool behavior
Resilience: retries, fallback behavior, state preservation, and graceful degradation
Usefulness: a real problem for a clear user
Demo clarity: show what failed, how the agent recovered, and why it worked
Prizes
First place: $3,000 cash + $1,000 TrueFoundry credits + $1,000 AWS Bedrock credits
Second place: $2,000 cash + $500 TrueFoundry credits + $500 AWS Bedrock credits
Third place: $1,000 cash + $500 TrueFoundry credits + $500 AWS Bedrock credits
Best social media prize: $1,000 cash for the post or thread with the most relevant impressions covering the hackathon and TrueFoundry AI Gateway.
Speakers
Preethi Kumaresan, AWS
Anant Agarwal, Principal Applied Scientist
Judges
Nikunj Bajaj, Co-founder and CEO, TrueFoundry
Preethi Kumaresan, AWS
[Additional judges TBA]
Who Should Join?
Builders working on agents that need reliability, fallback behavior, observability, safe tool access, and production-grade controls.
Join the Discord before kickoff so you get the submission link, support updates, and challenge announcements.
Useful Links
Sign up for TrueFoundry
Discord
Event calendar
YouTube
X
LinkedIn
About TrueFoundry
TrueFoundry helps teams build, deploy, and govern production AI applications. TrueFoundry AI Gateway provides a unified interface for accessing 1,000+ LLMs with routing, fallback, observability, and governance built in.