1 — Select your skills
How this demo works:
First you will see a baseline ranking — the top 10 jobs sorted by Euclidean distance between your skill vector and each job's skill requirements.
Rate those results; your votes are used later to compute the MRR but are not yet sent to the engine.
Then InnerMatch takes over: five rounds, two jobs per round, each round updating the model with your feedback before the next pair is shown.
At the end you can compare how well each approach aligned with your preferences.
The engine runs on AWS Marketplace via an online-learning API.
2 — Rate the baseline results
Baseline ranking — these 10 jobs are ordered by Euclidean distance between your skill vector and each job's required-skill vector (lower distance = better match).
This is a static, impersonal ranking: everyone with the same skills sees the same list.
Like or dislike as many as you want — your votes here are recorded only to measure preference alignment at the end via MRR.
They are not fed to InnerMatch yet.
3 — Guide InnerMatch
Adaptive learning — two jobs per round, two strategies:
Best match The job InnerMatch currently thinks fits your taste best, based on everything it has learned so far.
Fresh angle A job that is deliberately different from your pattern — shown so the engine can also learn what you don't want.
Your feedback updates the model before the next pair is selected. Five rounds, up to ten signals in total. The session is wiped from the server once you reach the results page.
Best match The job InnerMatch currently thinks fits your taste best, based on everything it has learned so far.
Fresh angle A job that is deliberately different from your pattern — shown so the engine can also learn what you don't want.
Your feedback updates the model before the next pair is selected. Five rounds, up to ten signals in total. The session is wiped from the server once you reach the results page.
Round 1 of 5
Fetching matches…
4 — Results: baseline vs adaptive
What you see here: The left column is the original Euclidean baseline — same static order as step 2.
The right column shows the exact 10 jobs InnerMatch picked across your 5 rounds, in the order they were shown
(round 1 → positions 1 & 2, round 2 → 3 & 4, …).
Green badges mark jobs you liked in each respective step.
The MRR (Mean Reciprocal Rank) below compares how well each approach surfaced your liked jobs early.