← Back to Projects

Market Pulse: Live Job Market Intelligence Platform

2026 - Present
Data Pipelines / MCP Server / SQL / Market Analytics / Live Deployment

Project Type

  • Association: Independent portfolio project
  • Role: Data Analyst / Data Engineer
  • Focus: Tracking real-time hiring trends for Data Analyst, BI Developer, Analytics Engineer, and Applied AI roles
  • Data Sources: Adzuna and RemoteOK job listing APIs, refreshed daily
  • Status: Fully built and live-deployed, running on a daily automated schedule

Objective

Built and deployed a live job market intelligence platform that combines a real-time data pipeline with a dual-transport MCP server layer. The pipeline classifies daily job postings from the Adzuna and RemoteOK APIs by role, seniority, and skills into a PostgreSQL and SQLAlchemy schema, running unattended via GitHub Actions and Docker on Render. On top of that foundation, a FastMCP-based MCP server exposes five job-market analytics tools over local stdio and remote streamable HTTP, secured with bearer-token authentication and per-IP rate limiting, with a Groq-powered LLM layered on top to translate the computed statistics into a plain-English market summary grounded strictly in the underlying data.

Tools & Technologies

PythonFastAPIPostgreSQLDockerFastMCPGroq API

Project Details

Engineered a PostgreSQL and SQLAlchemy schema to classify daily job postings from the Adzuna and RemoteOK APIs by role, seniority, and skills, turning scattered real hiring data into a structured, queryable foundation.

Automated the ingestion-to-aggregation cycle with GitHub Actions and containerized the service with Docker, deploying it on Render to run continuously without manual intervention.

Refactored shared query logic into a standalone module and built a dual-transport MCP server on top of it using FastMCP, exposing five job-market analytics tools over local stdio and remote streamable HTTP mounted on the existing FastAPI app.

Secured the remote HTTP transport with bearer-token authentication and a per-IP rate limiter, and layered a Groq-powered LLM summary on top of the computed statistics to translate raw hiring trends into a plain-English narrative grounded strictly in the underlying data.