AI Job Market Intelligence Platform

01Overview

This is an end-to-end data science product over 2025–2026 AI and data science job postings. It cleans and enriches raw listings, predicts salary and seniority, and turns the result into an interactive decision dashboard.

The questions are concrete: which roles are growing, which skills are in demand, how compensation shifts by location, experience and work mode, and which profile maps to a higher salary band. Model outputs are explained with SHAP, so the reasoning behind a prediction is visible too.

02Key Features

• End-to-end pipeline — from raw postings to a clean dataset, then to model and dashboard
• Defensive schema mapping — automatic mapping when column names differ
• Synthetic fallback data — a realistic dataset is generated when no raw data is present, so the app always runs
• Salary prediction — compensation bands and seniority levels predicted with XGBoost
• Explainability — SHAP shows how each feature moves a prediction
• Decision dashboard — interactive filters and visualisations in Streamlit
• Market signal — a layoffs/hiring signal included as market context

03Engineering Notes

• The product serves three audiences at once: recruiters benchmarking compensation, teams prioritising hiring, and candidates reading the market.
• MLflow and Optuna handle experiment tracking and hyperparameter search, and the synthetic fallback means the project runs on another machine even with no data.