MiMatch Data Science · AmbiMi's skills based job matching algorithm

The Data Science Behind MiMatch. Proprietary Data, Purpose Built AI, Proven Results.

The MiMatch Algorithm is AmbiMi's proprietary skills based job matching algorithm. Language models map every career to a proprietary skills taxonomy called the spine, the algorithm scores every match with explainable rules, and machine learning refines matching as Skill Hours™ accrue on every professional's AmbiFile.

The spine: a proprietary skills taxonomyReverse skills inferenceEvery match explainable
The Spine

A Proprietary Skills Taxonomy

At the core of MiMatch is the spine, AmbiMi's proprietary taxonomy of fields, job titles, skills and industries, produced by our data science team. MiMatch draws on thousands of real job matches studied over 20 years.

One title, one field

Canonical Job Titles

Every job is represented by one canonical title, and each title belongs to one field.

Objective and measurable

Precise Skills Profiles

Every canonical title carries its own precise skills profile, which powers skills matching across every field and industry.

Every way of writing it

Synonyms

Each canonical title carries its synonyms, so every way of writing a job leads to one clear title.

A second dimension

Industry

Carried by both the professional and the organization, it sharpens every match while the job title stays constant.

Sr Acct→Senior AccountantSupervisoryAccounting
Three Types of AI and One Proprietary Algorithm

Purpose Built AI, Each With One Job

MiMatch combines language models, a proprietary algorithm and machine learning, each with a distinct responsibility.

1. Large Language Models (LLMs)

Read resumes, career histories and job postings in any format, and gather the information that leads to the spine.

2. Small Language Models (SLMs)

Trained on the spine, they map that information to canonical titles, fields and skills with speed and precision at scale.

3. The MiMatch Algorithm

Scores every match with proprietary, explainable rules, factoring in every professional's Skill Hours™.

4. Machine Learning

Refines job matching as Skill Hours™ accrue, learning which matches lead to successful hires.

Reverse Skills Inference

Every Job in a Career History Is Data

LLMs read the titles and experience in each past job, and SLMs map each title to its canonical title in the spine and infer the skills applied in that job. The result is a measurable skills profile for every professional from day one, recorded on their AmbiFile.

Every Past Job

Skills Inferred

  • 1LLMs read the titles and experience in each past job
  • 2SLMs map each title to its canonical title in the spine
  • 3Result: the skills applied in every job, inferred
Every Job on the AmbiMi App

Skills Measured

  • 1Every skill used earns Skill Hours™
  • 2At 8 Skill Hours™, a skill becomes a verified skill
  • 3Result: verified skills, recorded on the AmbiFile
Skill Hours™: Measured Experience

Skill Hours™, Recorded on the AmbiFile

Every skill a professional uses on the AmbiMi app earns Skill Hours™. Skill Hours™ keep accruing, and every record is stored on the professional's AmbiFile, the only way to record these skills and carry them forward to future jobs.

8Verified at 8 Skill Hours™

Our data science set the verification standard: at 8 Skill Hours™, a skill becomes a verified skill. Verified skills and inferred skills feed the MiMatch Algorithm together.

Explainable Match Scoring

Every Match Has a Clear Reason Behind It

The MiMatch Algorithm scores every match with proprietary, explainable rules. Every search returns results, and the strongest matches rise to the top.

SignalWhat the MiMatch Algorithm Weighs
Job title and synonymsOne canonical title from the spine, however the job is written
SkillsThe precise skills profile of every canonical title
Skill Hours™Measured experience in every skill, including how it transfers to future jobs
SeniorityThe level of the role
IndustryCarried by both the professional and the organization
LocationWhere the job is, and where the professional is
PreferencesThe professional's job preferences
Proven Results

Built on Evidence, Sharper Over Time

Where it began

Built on Empirical Evidence

The MiMatch Algorithm began with thousands of real job matches studied over more than 20 years of operating 2 major recruitment agencies.

Always learning

Machine Learning That Grows With Skill Hours™

As professionals accrue Skill Hours™ on the AmbiMi app, machine learning learns which matches lead to successful hires and refines the MiMatch Algorithm over time.

Always

Designed for Ethical Job Matching

Built to remove hiring bias and designed to resolve the flaws in other systems. Every hiring decision is made by the organization.

Questions

Questions

What data shaped the MiMatch Algorithm?

Thousands of real job matches studied over more than 20 years of operating 2 major recruitment agencies. That empirical evidence was loaded into AmbiMi's data science before the MiMatch Algorithm was created.

What is the spine?

The spine is AmbiMi's proprietary taxonomy of fields, job titles, skills and industries. It trains AmbiMi's small language models and supports every match the MiMatch Algorithm makes.

What makes AmbiMi's skills taxonomy different?

The spine gives every job a precise, comparable skills profile built from objective, measurable skills. That keeps every canonical title unique and makes skills matching accurate across every field and industry.

What do LLMs, SLMs, the MiMatch Algorithm and machine learning each do?

LLMs read resumes and job postings and gather the information that leads to the spine. SLMs map that information to canonical titles, fields and skills. The MiMatch Algorithm scores every match with explainable rules, and machine learning refines matching as Skill Hours™ accrue.

Is the MiMatch Algorithm rules based or machine learning?

Both. The MiMatch Algorithm scores matches with proprietary, explainable rules, and machine learning refines job matching as Skill Hours™ accrue.

How is a skill verified?

A skill becomes a verified skill once a professional reaches 8 Skill Hours™ in that skill on the AmbiMi app. AmbiMi's data science team set that threshold.

Where are Skill Hours™ recorded?

On the professional's AmbiFile. Skill Hours™ keep accruing there, and the AmbiFile is the only way to record these skills and carry them forward to future jobs.

Proprietary Data, Purpose Built AI, Proven Results.

The spine, reverse skills inference and Skill Hours™ recorded on the AmbiFile give every professional a measurable skills profile, and every match a clear reason behind it.

Get the App