• Hackathon Radar
Hackathons
  • Browse
  • Judge Opportunities
  • Sponsors
  • Organizers
  • Map
  • Discover
Personal
  • Custom Views
  • Passport
  • Favorites
Explore
  • Stats
  • State of Hackathons
  • Changelog
  • Settings

Localhost: On-Device Agent Hackathon

lumaHosted on Luma

Fetched about 7 hours ago

Saturday, June 20, 2026

to Saturday, June 20, 2026

Artificial Intelligence

Event Type

in person

112

Participants

£500

Prize Pool

10

Est. Projects

Build an AI agent that runs entirely offline. No internet. No cloud. Just pure on-device intelligence. The Challenge Teams will build and deploy autonomous AI agents onto physical hardware, such as Raspberry Pi devices, similar edge hardware, or laptops, with zero internet connectivity. Your agent must operate completely self-contained; all processing, decision-making, and task execution happens on-device. What You’ll Build Using tools provided by our track partners and your own ingenuity, you’ll: Design an autonomous agent for a specific use caseIntegrate tools and capabilities within strict resource constraintsDeploy everything to run on limited hardwareDemonstrate your agent operating completely offline BYOD: Bring Your Own Device Participants are encouraged to bring their own devices if you have a cool setup you want to build on. This could include Raspberry Pis, edge devices, mini PCs! Deploying and running your agent fully offline on a laptop also counts as an on-device deployment, provided the agent operates without internet or cloud services during the demo! Track Partners Overmind — Automated end-to-end fine-tuning for autonomous AI agents. Overmind captures agent traces, identifies failure patterns, and continuously optimises model behaviour closing the loop from observation to improvement without manual intervention. Cosine — An AI coding agent trained for complex engineering work in real production codebases, with a focus on maintainable, reviewable code that fits existing architecture. Through their Lumen models, Cosine trains coding agents that pass complex tasks and are better for engineers to work with. Exo Labs — Run frontier AI models on everyday consumer hardware. Exo turns a cluster of devices into a distributed inference engine using pipeline parallelism. No cloud, no expensive GPUs. Run Llama, DeepSeek, and more completely offline. Cognee — An open-source AI memory engine that makes data structured, searchable, and production-ready for real-time intelligence. Cognee gives agents persistent, queryable memory so they can recall, reason, and improve across sessions without relying on cloud infrastructure. Captur — The first AI photo trust layer for enterprise that runs entirely on-device. Visual AI validates images in real time, in milliseconds, with no cloud lag. Models run locally; images never leave the device. Judges: Watch this space!! Prize Pool: £4000 🏆 Overmind - £500 Captur - £500 Cognee - £500 Cosine - £500 (+ £1500 in credits) ExoLab - TBC Zalos - £500 Schedule 09:00 – 09:30 Doors open, check-in & breakfast09:30 – 10:30 Welcome, opening remarks & track partner lightning pitches (Overmind, Cosine, Exo Labs, Cognee, Captur)10:30 – 11:00 Team formation & ideation11:00 Hacking begins13:00 – 14:00 Lunch (rolling — keep building)18:30 Submissions & Pizza18:30 – 19:15 Judging & deliberation (optional demos)19:15 – 19:45 Winners announced & prizes19:45 – 20:00 Networking & wrap-up Event Details 📅 Date: Saturday, June 20th, 2026 📍 Venue: Dawn Office, Level 7, Ilona Rose House, Manette St, London W1D 4AL 🚇 Transport: Tottenham Court Road (2 min walk) 👥 Capacity: 80 participants (teams of up to 4) Schedule TBCDawn T&C: https://drive.google.com/file/d/1vuI_bkZwzFyHYLtrrnO1r7CI5_u6WpRh/view

Judge Accessibility

Organizer email available25/25
Student-run event15/15
Actively looking for judges25/25
Small event (120 participants)10/10
No corporate sponsors10/10
New or emerging organizer10/10
Public registration available5/5
Online format (judge from anywhere)10/10

Top signals

Organizer email available
Student-run event
Actively looking for judges

Organizers

Alex Johnson

alex@example.org

Jamie Rivera

jamie@example.org

Sam Chen

sam@example.org

Estimated Audience

Mostly Students
ExperienceStudent
OccupationStudents
Beginner Friendly
Women in Tech

Technical Focus

AI95%
Web80%
Mobile25%

Industries

Healthcare
Education
Climate

Technologies

Python
React
OpenAI

Why this estimate

  • • Hosted by a university
  • • Open to students
  • • MLH member event

Estimate inferred from event metadata, not actual attendee data.

Quality Score

Quality Score

72/100
High confidence
Organiser16/20
Event Maturity14/20
Sponsors18/25
Participants12/20
Operations12/15

Why this score

Strong organiser track record
Returning event
Well-sponsored

Missing data

Prize details
Code of conduct