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
Shopping that keeps up when you change your mind. Stateful search across 50,000 products, offline by default, with optional LLM reranking.
Backend-enforced authorization that lets AI agents use private data without overstepping human intent or ownership.
Every coding agent gets a gate: changes become persistent only after trusted checks and a one-shot permit. Rejected attempts leave the workspace unchanged. No evidence, no effect.
EurekaML is An autonomous ML scientist that turns hypotheses into experiments, learns from every result, and evolves its own research strategy.
Fewer questions, better matches. Turning browsing into a conversation.
AI image detector
Handoff Gate: Trust every handoff, not every agent. It blocks unsupported claims, guides agents to correct them, and keeps unreliable research out of the final report.
Coconut House optimises kernel for transformer layer for an Nvidia Geforce RTX 3060 Laptop
An offline, CPU-only shopping copilot that remembers user preferences across turns, asks clarifying questions, and reranks 50K products with 99% HitRate@10 on the public set.
We built a conversational shopping copilot that understands changing intent and constraints, achieving 100% Hit Rate@10 and a 0.965 technical score, about 9x the baseline.
An offline shopping copilot that understands natural intent, adapts across turns, asks smarter questions, and ranks products transparently without APIs, GPUs, or runtime cost.
A lightweight conversational shopping prototype that tracks changing preferences and returns catalog-grounded recommendations with a local, no-LLM pipeline.
Say it once. Threadline listens, remembers, and adjusts as your mind changes, so shopping feels like a conversation, not a search bar. No LLM, no external API cost, just the right product, faster.
Other platforms trust the agent, then check the damage. Heimdall decides what the agent can do before its container starts. The container never holds the key.
AI images are everywhere nowadays. Our detector checks if an image is synthetic, whether it be compressed, blurred, resized or edited, so people can question what they see, not blindly trust it.
Harness buddy automates all the ML research for you. It writes training code, reads its own crashes, and critiques itself when stuck. Beat the FM baseline on hidden test. Zero human interventions.
BranchPoint enables safe multi-agent collaboration with isolated workspaces, persistent branches, recoverable checkpoints, security checks, and controlled merging.
Needle in a haystack, solved by conversation
Scoutie separates the decision from the conversation. Instead of hiding decisions inside an opaque LLM, Scoutie's decision engine uses information theory to determine the most useful question next.
KernelKraft is an adaptive Transformer inference accelerator that delivers up to 1.471× end-to-end speedup—with a 1.272× geometric mean across 13 official workloads—on an NVIDIA T4.
Ask smarter, not more: a shopping agent that finds one hidden product among 50,000 in ~3 turns. Achieves 82% technical score on a public data set, offline with zero LLM calls.
Agent Launchpad gives AI agents scoped, revocable keycard-style tokens enforced at the backend — so prompt-injected overreach gets denied, every action is traced, and access can be revoked instantly.
A conversational e-commerce search agent, touching lives, making shopping easier.
An autonomous LLM agent that runs the full ML research loop end to end on a recommender benchmark: EDA, literature and web search, hypothesis, code, train, evaluate, reflect. No human in the loop.
433 – 456 of 614
