Skip to main content

What a cohort is

A cohort is a group of 4–8 learners matched by the platform. Your cohort peers are on the same career track, at a similar progress level, and strong in different skills than you — so you can learn from each other. Cohorts replicate the experience of working on a real team: accountability, peer feedback, and the kind of learning that only happens when you have to explain something to someone else.

How matching works

The algorithm runs every Sunday at midnight UTC. It:
  1. Builds a skill mastery vector for each unmatched learner
  2. Scores every possible pair: 40% similar progress + 60% complementary skills
  3. Groups learners into cohorts of 4–8 using those scores
You receive a notification when you’re matched.

What your cohort does together

Weekly challenges

Every Monday, your cohort gets a new coding challenge for your track. Submit your solution by Sunday. Examples:
  • AI Engineer: “Build a Mini RAG Pipeline — scored on retrieval quality and code review”
  • Data Scientist: “Model Bake-Off — best F1 score on the holdout set wins”
  • ML Engineer: “Optimise a slow PyTorch training loop by 3x”
  • Data Engineer: “Build a dbt project with staging, intermediate, and mart layers”
+150 XP for submitting. Bonus XP for top score.

Monthly competition

Once a month, a larger capstone-style challenge judged by your peers. +500 XP for the winner.

Peer code reviews

When you submit a lab, a cohort peer is automatically assigned to review your code within 48 hours. Reviews must include at least 20 characters of written feedback and a score out of 100. +50 XP for completing a review.

Standup feed

A simple async feed for daily updates. Even one sentence keeps the cohort active and earns streak credit.

Weekly leaderboard

Weekly XP ranking within your cohort — resets every Monday.

Cohort health score

Each cohort is scored weekly from 0–100:
ComponentWeight
Standup post rate30%
Peer review completion rate40%
Challenge submission rate20%
Active member rate10%
A score below 40 flags the cohort for admin review.