At Hack the North, there's no first, second, or third place in our main award — top projects become finalists.

We love projects that are creative and surprise us: original ideas, clever technical work, and creative experiences that make people excited about what's possible. Your project can be playful, experimental, wonderfully weird, or useful. It doesn't need to become a real startup with a business plan or solve major world problems! Follow your curiosity, try something new, and build something you're excited to share, and you'll do great!

To see past finalist projects from Hack the North and get an idea of what makes a great project, check out museum.hackthenorth.com!

The prizes for becoming a finalist are a surprise!

Use the OpenAI API to build something ambitious, with Codex as your development teammate. Whether Codex helps you ship a focused feature, turn a prototype into a polished product, or coordinate an end-to-end build, show us how it helped you go further than you otherwise could.

Projects will be assessed on:
- What you built with the OpenAI API: how creatively and effectively the API powers the experience.
- How Codex helped you build it: how meaningfully it supported planning, implementation, testing, debugging, or iteration.

During your demo, show the working product, explain how the OpenAI API is used, and share one concrete way Codex improved your process or outcome. First, second, and third place will go to the strongest projects across both areas.

Check out our documentation on Codex use case examples (https://learn.chatgpt.com/use-cases?task_type=code) to get started, and feel free to stop by our booth if you have any questions!

Prizes:
1st Place - Lunch or Dinner with OpenAI Recruiters & Engineers, ChatGPT Pro Subscriptions, OpenAI Swag Bundle
2nd Place - ChatGPT Pro Subscriptions, OpenAI Swag Bundle
3rd Place - OpenAI Backpack & T-Shirt

Financial research arrives as sprawling reports, transcripts, and filings—too large for any single context window. You'll design a custom agent/harness that ingests massive, real-world research data via an MCP tool and answers precise, citable questions accurately, even when the corpus is noisy or contradictory.

Phase 1: Master the known dataset. Ingest it, structure it, query it—proving accuracy, precision, and auditability at every step.

Phase 2: Can your system generalize? On judging day, a completely new industry dataset appears live. Your harness must adapt immediately—no pre-built solutions, no shortcuts.

This challenge rewards systems that genuinely work: smart retrieval, reliable grounding, and honest handling of ambiguity. The challenge isn't prompt engineering. It's real infrastructure under real pressure.

At Shopify, we're obsessed with making commerce better for everyone—merchants and the customers they serve. Your mission: experiment where AI meets commerce. Build something that makes us say "wait, that's possible?" We're looking for technical innovations that challenge what AI can do for entrepreneurs—from foundation-shaking algorithms to delightfully practical tools. The only constraints? Your imagination and the hackathon clock.

Optional: Tap into Shopify's ecosystem to accelerate your build—Shopify API, Remix, Polaris design system, and more. Take inspo from one of our projects - SimGym.

Theme ideas (not requirements):
- AI-powered merchant superpowers
- Commerce intelligence & data insights
- Workflow automation
- LLM applications in the commerce ecosystem
- Customer experience enhancements

Judging Criteria
- Technical Excellence: Sophisticated, appropriate use of AI with quality execution
- Impact Potential: Meaningfully improves merchant operations or customer experience
- Innovation Factor: Unexpected approaches that make us think differently about commerce

Prize: Winning team receives (per person): Shop Cash!

Build a hack that focuses on improving the developer experience in some meaningful way - this could be tackling any part of the development lifecycle (creating, modifying or testing software)!

Prize: Keychron keyboards for each member of the team.

Eligibility
No API requirements for this prize! You do not need to use Warp, but we'd love for you to try it out to build this + check out the Oz API!

Judging Criteria
Does the project improve the developer experience?
- Wow factor: Does the project stand out?
- Technical difficulty: How technically complex is the project?
- Originality: Is the project new or unique?
- Design: Is the project easy to use and polished?

Note that many developer tools may not have traditional GUIs - we are referring to all relevant aspects of UI/UX design here, there's no need to add a GUI for the sake of adding one.

We are looking for ambitious projects that protect the quality and authenticity of our world's information, using our AI detection and hallucination detection APIs. Firstly, we are interested in investigative projects that use our API to scan context and find AI slop and hallucinations from noteworthy, unexpected, or significant sources. Secondly, we are also interested in projects that integrate AI or hallucination detection in significant ways, to create products or platforms the world sorely needs in an age of rampant AI slop.

Prizes:
1st place - AirPods Pro 3 + Premium GPTZero swag + $2000 of GPTZero credits per teammate
2nd place - Premium GPTZero swag + $500 of GPTZero credits per teammate

Create an embedded system with QNX that uses AI!

Hard requirements:
- Project uses QNX OS
- Project includes one of the open-source AI modules available at https://oss.qnx.com

Judging criteria:
- Is it a "cannot-fail" embedded application?
- Does it require real-time or reliability?
- Is it using AI in an interesting or creative way?
- Is it running on the embedded hardware or QNX VM (and not in the cloud)?

Dryft is building the future of manufacturing intelligence.

We're giving every team an autoregressive transformer and one goal: make it generate tokens as fast as possible.

How you get there is up to you: write Triton kernels, fuse operations, rethink memory layouts, or build an autoresearch agent that experiments, benchmarks, and discovers optimizations for you.

Submissions will be run on H100s and must pass correctness tests before being benchmarked on hidden workloads.

The team with the fastest and correct implementation wins $2000.

Read more: htn.dryft.ai

We're looking for projects that take genuinely messy, unstructured, real-world data (transcripts, logs, PDFs, scraped forums, sensor streams, multilingual text and more) and turn it into something a person or agent can actually act on. The most exciting projects will build agentic systems: agents that can reason over your data, decide what to retrieve, call tools, and trigger actions on their own, with Elasticsearch as their context layer. That means going beyond RAG with a chatbot. Think hybrid search combining BM25, Jina dense vectors, and reranking, plus aggregations, ES|QL, geo or time-series queries, with Agent Builder tools and Workflows that close the loop by taking action, not just answering questions.

Prizes:
1st place: Meta Quest 3S VR Headset
2nd place: Bose QuietComfort Headphones

Baseten is a leading platform for both inference and training, and we want you to come up with a creative way to incorporate it into your hack! Check out our GitHub repo to get started, and if you have any questions, come visit our booth!

https://github.com/basetenlabs/Hack-the-North-2026

Prizes (per team member):
Grand Prize: Free trip to SF to meet the Baseten team + final round interviews with Baseten + $200 of Baseten credits + signed hardcover of Inference Engineering
Runner-Up: AirPods 4 + $100 of Baseten credits + signed hardcover of Inference Engineering

To qualify for this award, your team must build a system that leverages large language models to operate on real-world, messy data and take meaningful actions. More specifically, qualifying teams must build agents that can handle unstructured information, incomplete datasets, conflicting sources, or noisy data - the kind of messy reality that exists in customer databases, support tickets, financial records, or operational systems. Teams should demonstrate techniques like data cleaning and validation, multi-source resolution, intelligent error handling, or robust decision-making under uncertainty.

Teams will be judged on technical complexity, creativity, how well the system handles real-world messiness, and practical utility. Show us agents that don't just work in clean demos, but can actually operate in the chaotic data environments that businesses deal with every day. We will award 2 prizes: $10K for top and $2K for second place.

Here at Big Company Incorporated Inc., we take cybersecurity very seriously. That's why we'd like you to review some of our internal traffic logs and make sure there's no funny business going on. If you find anything, let us know the who, what, when, and how. Although we're very sure that our current threat detection model, which is to trust that everyone is just going about their assigned business, is completely effective!

... but just in case it isn't, maybe having a more robust detection solution would be a good idea. Can you also put something together that will look at this data as it comes in, and identify any weirdness? Maybe so we don't have to ask a bunch of hackers to comb through our log data again, cough.

(See our Discord channel for the dataset to analyse, and don't be afraid to come chat with us for hints!)

Take the foundation from hello-pear-qvac-tui and turn it into something uniquely yours. Your project must use the repository as a core part of the application, then push it further to create a local AI experience that is innovative, useful, and genuinely cool.

Build anything from a powerful developer tool to an unexpected everyday application—but make it your own. Projects will be judged on innovation, usefulness, technical creativity, and overall wow factor.

Two prizes (1,000 USD and 500 USD) will be awarded to the projects that best demonstrate what's possible when you build on QVAC and Pear.

Prize Description: $100 gift cards for a team of 4 + an exclusive opportunity to interview with our tech and lab team

Eligibility: Creating modern and new ways to get car and tenant insurance via AI (chatbots, AI agents, etc.)

Judging Criteria:
1. Build a functional prototype that reimagines how people obtain car insurance, tenant insurance, or both.
2. Demonstrate a complete or partial quoting experience, allowing a user to provide relevant information and receive an insurance recommendation, estimate, or next step.
3. Prioritize user experience and accessibility, with a clear, intuitive, and inclusive interface.
4. Provide a brief explanation or README describing: the problem being solved; how AI is used; the user journey and key features; any assumptions, limitations, or future improvements.
5. Ensure that the project was created during the hackathon and is substantially the team's own work.

WHITEOUT drops hackers into a live Arctic simulation. Your team will write the shared intelligence that coordinates a heterogeneous fleet — fixed-wing aircraft, quadcopters, rovers, and fixed sensor towers, all running ArduPilot and MAVLink — to detect, classify, and track a moving target across contested terrain. No swarm experience required; bring your best distributed thinking.

Teams are scored live on coverage, collaboration, efficiency, and tracking accuracy. The best agents earn the API Prize.

1st place walks away with $2,000, 2nd place $1,000, and 3rd place $500. 1st, 2nd, and 3rd place also receive a guaranteed first-round interview with Dominion Dynamics in addition to the cash prizes.

Build a mobile app with Expo, the open-source framework for native iOS and Android apps with React Native. We're looking for the best mobile experience: an app that's beautiful, feels truly native, and is a joy to use. Use whatever parts of Expo get you there - Expo Router for navigation, Expo UI for native SwiftUI and Jetpack Compose components, expo-widgets to add widgets and Live Activities to your iPhone, and so much more (see docs.expo.dev).

Prize: $150 & an Expo swag box for each teammate

Build a creative, useful, or technically ambitious project that uses Bracket Bot to interact with the physical world. Use navigation, voice, manipulation, teleoperation, robot learning, or surprise us with something completely new. We're looking for a sick demo where the robot is essential to the idea.

Judging: Meaningful use of Bracket Bot, technical execution, creativity, impact, and demo quality.

Prizes:
1st: 4x Bambu Lab A1 Mini + 4x SO-101 leader-follower kits
2nd: 4x SO-101 leader-follower kits
3rd: 4x SO-101 follower robots

Most teams stop at catching errors. We want to see you go further. Build with Sentry and use at least two products beyond error monitoring: Session Replay, Logs, Tracing, Profiling, Uptime Monitoring, or our MCP/AI agent monitoring, to debug, understand, and improve your hack. Show us how observability actually shaped what you built: the slow endpoint you found in a trace, the broken flow you caught in a replay, the bug you squashed at 4am because your logs told you where to look.

We'll judge on creativity, depth of integration, and how meaningfully Sentry data influenced your project, not just whether the SDK is installed.

Prize: every member of the winning team gets a guaranteed interview with the Sentry team for our internship or new grad roles. Skip the resume pile entirely.

The prize will be awarded to a team whose members have collectively demonstrated excellence and where all team members have attended one (1) or fewer hackathons prior to Hack the North 2026.

All eligible teams meeting this criterion will be eligible to be considered for this prize.

Our SDK gives you full control over LeLamp's character and behavior. We're looking for the most creative or technically impressive use of our robot platform, whether that's genuinely useful, delightful, or just weird in the best way. Docs and example code will be provided at the event.

Prize: 1 Bambu A1 printer and 1 LeLamp (total per team)

Build anything. The only rule is it runs on Backboard.

Text, image, voice, and 17,000+ models on top of state management, RAG, memory, embeddings, tool calling, web search, and adaptive context management, all through one API. The glue is gone. The only limit is how far you push the idea in 36 hours.

We judge ambition. Not polish, not pitch decks. We want the project that makes the room say "you built that this weekend?" The more of the stack you use, the crazier it gets, the better your odds.

Prize: Guaranteed internship interviews, $400 (per team), swag packs

AI agents are moving beyond single-agent assistants toward teams of specialized agents that collaborate to solve complex tasks. This challenge asks hackers to build a functional multi-agent application that solves a meaningful real-world problem through genuine agent collaboration — task decomposition, communication, tool use, and coordination — rather than simply chaining LLM prompts together. Teams are encouraged to build on JiuwenSwarm or its successor, WorkSwarm, though this isn't required. Teams will be judged on quality of multi-agent collaboration, scenario creativity, demo completeness, technical implementation, and reusability.

We will award 2 prizes: 1st place receives a Huawei Watch GT 6 + an internship opportunity with the openJiuwen team at the Huawei Toronto Research Center, subject to interview (per team member). 2nd place receives Huawei FreeClip 2 Earbuds + an exclusive office tour (per team member). Each participating team may also receive up to $40 USD in API credits.

Full challenge details, sponsored API credits, and resources: https://github.com/openJiuwen-ai/jiuwenswarm

AI is moving beyond text-based chat toward real-time, multimodal interaction. OMNI Live challenges hackers to build a functional AI application for a real-world edge-device scenario - a phone, wearable, smart home device, robot, or vehicle - using an OMNI multimodal model accessed via cloud APIs. Qualifying teams must meaningfully incorporate all three core modalities (vision/video, speech/audio, and language) and demonstrate at least one complete end-to-end scenario; concept-only designs and static mockups won't be considered. Teams will be judged on scenario creativity, effective use of OMNI's multimodal capabilities, demo completeness, interaction quality, and technical implementation.

We will award 2 prizes: 1st place receives a Huawei Watch GT 6 + an exclusive office tour at the Huawei Waterloo Research Center (per team member). 2nd place receives Huawei FreeClip 2 Earbuds + the same office tour (per team member).

Full challenge details, sponsored API credits, and resources: https://github.com/cari-waterloo-rc/OMNI-Live-Build-the-Next-Generation-of-Real-Time-Multimodal-AI