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
Sort by:
Subtitulado y traducción simultánea para conferencias técnicas con Gemini Live, rotación continua sin cortes y respaldo local 100% offline.
A 49.3M parameter language model trained from scratch with intelligent data filtering, achieving 66.6% loss improvement and coherent text generation under strict parameter limits.
An OCaml → CUDA compiler with a built-in second opinion.
It's noise cancellation but it's intelligent
In high acuity fields such as autonomous vehicles
An AI in games that really feels like an AI
a compiler to discover non-trivial optimizations
Blazing fast LLM inference.
An agent loop that searches fusions, layouts and launch structure trying to approach the memory bandwidth floor.
Vigil is your AI powered security guard. Designed to catch anomaly's that are easy to miss.
Agent for adapting, optimizing, and running code on any machine, anywhere.
Charlie is Openrouter for robotics. We reduced control error by up to 123x compared to a pretrained GR00t-N1, a general purpose robotics model, and outperformed Tra-MoE by up to 24% on average.
Blackwell turns surgical papers, textbooks and lecture videos into a physics-accurate VR practice room, with every cut line traced back to the exact page or timestamp it came from.
BigBird ends the choice between complicated pro software and limited AI tools, combining cutting-edge AI alongside human creativity
Turn idle GPUs into instant, distributed compute.
A small language model built from scratch to make code smarter.
FAULTLINE AI detects LLM bugs, hallucinations, and security risks, explains failures, and recommends fixes to help developers build safer, more reliable AI applications.
A language model that fits under 50M parameters and under your desk. Trained from random noise on one RTX 4070 in 4h22m. No pretrained weights. No teacher. No API key.
An OpenUSD visual schema tracking artwork genealogy and 6-axis semantic scoring, powered by NVIDIA-accelerated gondurastration neural synthesis.
QuantGate reviews quantized robot policies before they ship to a Jetson: cost on the device, blinded real-arm trials, and an honest verdict. Earn the next test, not the approval.
A 42M-parameter looped Transformer trained from scratch on 5B tokens with a single consumer GPU.
Physics-powered AI that predicts geomagnetic threats and protects critical power transformers before failure.
RoadLens AI turns ordinary cameras into traffic intelligence by detecting and tracking vehicles, identifying speeding, reading plates when possible, and creating reviewable event reports.
Baseline turns any tennis match into a Learner’s Feed—an AI coach explains the strategy behind every point and answers questions while you watch.
