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
PhytoBridge links traditional medicinal plant knowledge to phytochemical and published scientific evidence, so anyone can see what tradition claims and what science actually supports, side by side.
An offline-first DSA streak companion: daily problem picks, a GitHub-style heat grid, and spaced revision that resurfaces solved problems. Everything lives in your browser, nothing leaves your device.
Split a secret key into 5 Shamir shares; any 3 rebuild it for an agent task, then zeroize. The agent never holds the key.
An AI agent that answers questions over encrypted files. Keys are Shamir-split into 3 shards and never exist whole: decryption happens in RAM only.
A voice-first AI bridge in Marathi and Hindi: anyone without a smartphone calls in with a civic, health or scheme problem and the AI routes it to the right office or local volunteer.
A backwards exam-revision planner: enter your syllabus and exam date, get a day-by-day plan, and re-cut it in one click when you slip a day.
A working front-end prototype where a real, deterministic policy engine scores every AI agent tool call on screen: verdicts, reason codes, and an audit trail you can change live.
Samajh reads confusing Indian documents and government forms and explains them in plain language in ten Indian languages, powered by Google's Gemma 4.
A review console that checks clinicians' planned orders and actions against the patient's record and flags risky or conflicting steps before they're signed off.
A trust review dashboard where a human approves, rejects, or audits every high-stakes action an autonomous agent proposes, before it executes.
A trust review console for AI agents: every payment, message and CRM write waits for a human verdict, with a full audit trail.
A review console that checks proposed clinical actions against patient context and audit rules before they execute, helping catch unsafe or unnecessary orders before they reach a patient.
Small businesses pay vendors by hand. vendor-payment-system automates it: an agent drafts the payout run from the ledger, the operator approves line by line, PayPal executes in sandbox.
The human side of clinical AI safety: every agent action waits for a human look.
The human side of agent safety: consequential AI actions land in an approval queue with the evidence a human needs to decide in seconds.
Climate-data agents read untrusted sensor feeds and public reports. Seven typed questions gate every tool call: PROCEED, CONFIRM for a human, or DENY.
A clinical agent that reads patient documents is the highest-stakes prompt-injection target there is. Every tool call passes seven typed questions; anything risky holds for a clinician.
Agents handle logistics beautifully and taste terribly. AOS_engine gets taste: a Qloo-backed get_affinity tool filters venues, gifts and outreach by cultural affinity, not distance.
