Devpost
Participate in our public hackathons
Devpost for Teams
Access your company's private hackathons
Grow your developer ecosystem and promote your platform
Drive innovation, collaboration, and retention within your organization
By use case
Blog
Insights into hackathon planning and participation
Customer stories
Inspiration from peers and other industry leaders
Planning guides
Best practices for planning online and in-person hackathons
Webinars & events
Upcoming events and on-demand recordings
Help desk
Common questions and support documentation
A shopping agent that finds a hidden product in 2 turns, by computing which question reveals the most — and refusing to answer until it knows. $0, offline.
Buyte is an AI shopping copilot that remembers your preferences, asks the right questions, and narrows 50,000 products down to find what you actually want by understanding and adapting to you.
Shopping copilot that finds the exact product in ~2 turns — TechnicalScore 0.9715 (9× baseline), $0 and 0 LLM tokens on the scored path. Learned + LLM layers handle real shopper wording.
Detects AI-generated images and holds accuracy after real-world compression, blur, crop & rescale using a hybrid CLIP + FFT-forensics model. 0.996 AUC clean, 0.984 across 15 transforms.
A shopping copilot that inverts the shopper instead of matching keywords: exact-posterior retrieval plus information-gain questions. Hit@10 1.000, MRR 0.984, 1.96 turns, $0 per session.
TraceLens is ByteCop’s robust AI-generated image detector, built to remain reliable after real-world transformations and scale efficiently for platform use.
It controls what an agent is allowed to know, proves what it actually knew, and repairs the damage when a fact turns out to be wrong.
An AI agent that dynamically analyses, learns, and recommends products to users using multiple layers of item identification, ranking by relevance, and asking the user for clarification when needed.
In today's day and age, can you still tell the difference between real and AI-generated images? We strive to build a robust AI-image detector that survives compression, cropping and filters
A transformation-aware AI image detector that combines semantic and forensic evidence to remain reliable after compression, resizing, and real-world image transformations.
Ever wanted a personal intern that constantly works? Just hire me! Or you can use this AI scientist that I made.
Shop Shop: Your Favourite Shopping Companion - Get What You Need ASAP! is a fully offline conversational agent that is built for E-Commerce Shopping.
0.9746 on the official evaluator, running offline with zero tokens. A second adversarial benchmark revealed the same agent scores 0.348 against a customer who paraphrases. Both numbers reported.
Spot the synthetic. Protect the authentic.
An autonomous ML research agent for recommender systems. It forms hypotheses, patches its own code, executes iterative optimization loops, and verifies results via bootstrap on sealed splits.
Most AI-image detectors work great in the lab but fall apart the moment an image is compressed or blurred. We built a dual model that holds up and stays fairly accurate at 85% or higher.
DINOv2 AI image detector tested on 19 real-world corruptions, not just clean data. Caught and fixed a data shortcut causing a fake score. Added FFT forensics, raised worst-case AUROC from 0.82 to 0.88
Team CAPybara knows when someone's capping. So with our skillset, is it any surprise we went for Track 5? With our solution, you can clock what's real, because getting fooled by an AI? Now thats an L.
Volc Agent Launchpad is a three-layer middleware that lets safe AI coding work pass, contains dangerous prompts and reported runtime actions, and exports verifiable trace evidence.
Zero humans in the loop: an LLM agent that runs its own recommender-systems research cycle — hypothesis, implementation, leakage review, measurement — and beat the official KuaiRand baseline.
A shopping agent that shows one product instead of ten when it is unsure. Hunch's confidence is a measured quantity, AUC 0.845. 50,000 items to the right one in 2.29 turns, offline, zero dependencies.
CRIS bridges user dialogue and e-commerce discovery across a 50000 product backlog through a combination of Dynamic State Tracking, Hybrid Retrieval and information entropy calculations.
A shopping agent that decides, every turn, whether to answer you or ask one more question. Solves all 200 test sessions in 2.8 turns — 9x the official baseline, on zero tokens and no network.
Reflex is an autonomous ML research agent that runs the entire model-improvement loop by itself. Writing code, training, evaluating, and iterating until it converges on a optimal solution.
385 – 408 of 614
