Projects / Hack the North 2025

FinalistHack the North 2025: FinalistsConfidence: medium

SpeakEasy

Behavioral-interview practice that scores what you say and how you say it from video.

Original submission ↗Use as a reference →
Project facts
Event
Hack the North 2025
12–14 Sep 2025
Team
Size unknown
Build
software only
AI
llm api, llm scoring, computer vision, voice ai

Problem

Video interview analysis tools are sold to recruiters; applicants only meet them in the real interview and have no fair way to practice against that kind of scoring.

Solution

Pick behavioral topics, answer generated questions on camera, and get verbal scores (relevance, STAR structure, insight, vocabulary, filler words) plus non-verbal scores (face, eyes, gestures, pauses, framing) with concrete pros and cons.

Demo moment

Record one answer and immediately see a breakdown of both your content and your body language.

Why it stood out

  • team statedInverts a crowded enterprise market: the same analysis recruiters buy, pointed at the applicant.Source ↗
  • team statedNamed a list of incumbents in the write-up, showing the team knew the market it was flipping.Source ↗
  • team statedEngineering discipline on LLM output: benchmarked providers, then added validation with automatic re-prompting until output met format rules.Source ↗
  • curator inferenceUsed a sponsor's model router as core infrastructure, fitting a sponsor prize lane alongside the main rubric.

Lessons for your next hackathon

  • Look for a tool that exists only for the powerful side of a transaction and build the other side's version.

Similar situations

Projects that share domains, mechanisms or judging lenses — often with a different problem and stack.Tune the comparison.

Award winner · 2026

Multi-cluster AI inventory controller

Kubernetes-native controller that inventories AI infrastructure across clusters, inspired by agentregistry.

Shared: mechanisms sponsor platform mastery, visible engineering depth · lenses technical depth, sponsor api use

Finalist · 2025

DUM-E

Voice-commanded robotic arm that sees its workspace and acts on natural-language instructions.

Shared: mechanisms visible engineering depth, one sentence clarity · lenses technical depth · ai computer vision, llm api, voice ai

Runner-up · 2026

Kubernetes-native clinical MCP federation

Clinical AI agent platform that federates several MCP servers behind one endpoint to query MIMIC-IV securely.

Shared: mechanisms sponsor platform mastery, visible engineering depth · lenses technical depth, impact

Grand winner · 2024

FurMe

A Furby-shaped desk assistant that follows your face, reads your mood and talks back.

Shared: mechanisms one sentence clarity, sponsor platform mastery · ai llm api, computer vision, voice ai

Award winner · 2026

Kubernetes-native MCP governance

Monitors agentgateway and kagent resources, scores MCP security posture and adds AI risk analysis.

Shared: mechanisms sponsor platform mastery · lenses technical depth, impact · ai llm api

Sources

  1. https://devpost.com/software/speakeasy-vdkc9n · devpost project · retrieved 2026-09-14