Skip to main content

How it Works

Transparency First

How the matching
actually works.

No black boxes. No buzzwords. Here's exactly what happens when you post a job or create a profile on SeaPlatformJob.

First: what our AI does — and what it doesn't.

✓ What it does well

Converts unstructured CV text and job descriptions into comparable skill profiles. Scores candidates on maritime-specific criteria (certifications, vessel type, fuel systems). Factors in working style and culture preferences. Ranks and explains its reasoning.

✗ What it can't do (yet)

It cannot predict performance. It cannot replace a good interview. It doesn't know if someone is reliable or a team player — only you can judge that. It improves with every rated placement, but right now it has limited data. We'll tell you when that changes.

The baseline it beats

Traditional keyword search, which misses qualified candidates with different vocabulary. LinkedIn's general algorithm, which doesn't understand STCW, vessel types, or maritime culture. Gut-feel hiring, which costs the industry billions annually in bad fits.

How it improves

After every placement, both the employer and the candidate rate the match quality. This feedback is anonymous and feeds directly into the model. The more placements we make, the better the ranking becomes. We publish these scores publicly every month.

Step by step

From signup to placement — here's exactly what happens.

01

Build a structured profile — not just a CV upload

When you sign up, you don't just upload a PDF and disappear. Both candidates and companies complete a structured profile that captures what traditional CVs miss.

For Candidates Certificates & licences (STCW, MCA, class endorsements), vessel type experience, fuel system familiarity, preferred work culture, career ambition direction, and a 10-question working style assessment.
For Companies Fleet type & size, upcoming vessel technologies, team culture descriptors, hiring urgency, and the working style profile of the ideal candidate — not just their qualifications.
02

The matching engine builds compatibility vectors

Every profile is converted into a mathematical representation — a "vector" — that captures skills, experience, and personality dimensions. This happens automatically using language models trained on maritime data.

What gets vectorised Hard skills (certifications, vessel types), soft skills (leadership preference, adaptability signals), culture fit dimensions (startup vs corporate, structured vs autonomous), and green maritime expertise (LNG, methanol, CII, EU ETS knowledge).
03

Matches are ranked — with explanations

You don't get a list of 200 applicants. You get a ranked shortlist of the top matches, with a plain-language explanation of why each candidate was ranked where they were.

Example explanation "Anna K. ranks 1st because her LNG certification and dual-fuel vessel experience directly match your requirements. Her culture profile aligns strongly with your 'structured, safety-first' preference. One gap: she has no experience with your specific vessel class — this can be discussed at interview."
04

You decide — the AI advises, not replaces

After reviewing matches, you initiate contact through the platform. After a placement is made, both sides rate the match quality (1–5 stars, plus qualitative feedback). This data improves every future match on the platform.

Our commitment We publish monthly match quality scores from placed candidates. If the algorithm underperforms, you'll see it — and so will we.

Under the hood

For those who want to know what technology is actually running this.

Profile Processing

NLP & Entity Extraction

Job descriptions and CVs are processed using large language models to extract structured data: certifications, vessel types, skills, and experience duration. No manual tagging required.

Matching Engine

Vector Similarity Search

Candidate and job profiles are embedded as high-dimensional vectors. Similarity is computed across skill, culture, and experience dimensions. Ranking weights are tuned by placement outcome data.

Compatibility Layer

10-Question Psychometric Snapshot

Not a full personality test — a short, validated questionnaire that captures working style preferences. Results are converted to vectors and included in the match calculation, weighted at ~20%.

What we don't claim

We prefer honesty over a sales pitch.

We don't claim to eliminate bad hires

We reduce the risk of skill mismatch and culture misalignment. Whether a hire works out still depends on management, onboarding, and many factors we can't see.

We don't have years of data yet

The algorithm improves with placements. Right now we have early data. We'll tell you when the dataset grows large enough to make statistically robust claims.

We're not a replacement for interviews

Our job is to get the right shortlist in front of you faster. The interview and final decision are yours — and should be.

We don't know everything about maritime

Certain niche vessel types, regional manning conventions, and flag state specifics are areas where our data is thin. We flag this when it affects a match.

Early placement results

4.3
Avg Match Rating
out of 5, rated by both sides
11
Days to First Match
median, first cohort
78%
Would Use Again
employer post-placement survey
34
Placements Made
as of Feb 2026 · growing weekly

These numbers are from our first cohort of placements. We update them monthly. View full transparency report →

Ready to see it in practice?

Post your first job free, or create a candidate profile in under 10 minutes.

SPJ is not just a platform; it's a transformative force in the maritime sector. We reinvent job discovery and collaboration, leveraging cutting-edge AI to create a space where careers thrive and innovations set sail.

Featured Posts