IKARO, AI hiring
AI recruiting: an agent screens each candidate's skills, preferences and conditions, then LLM comparisons rank them per role.

The problem
Reviewing hundreds of CVs per role is slow and depends on who reads them.
Keyword filters drop good candidates who say it in different words.
What I built
A semantic-similarity prefilter that narrows the list, and three AI judges that score different dimensions of fit in parallel.
A weighted score with each judge's explanation, so the recruiter sees why someone ranked high.
How
Vector prefilter
pgvector narrows thousands of profiles to the ones closest to the role.
Parallel judges
Each judge looks at a different dimension and writes down its reason.
Explainable score
The shortlist can be audited candidate by candidate.
Got a similar process?
Tell me about it. The first call is free.