← All projects

Automation Platform

AI Job Search Agent

A local-first job discovery and application workflow that collects opportunities, ranks fit, generates tailored materials, and stages browser-based applications for review.

PythonFastAPISQLiteSQLModelOllamaPlaywrightAPSchedulerJinja2
Active
5,657jobs processed
15FastAPI endpoints
885generated artifacts
6source integrations

01 · Problem

What the system solves

Job searching is fragmented across boards and repetitive at every stage. The agent consolidates ingestion, relevance scoring, document generation, review, and application staging into one workflow.

02 · Architecture

How the pieces connect

01Job sources
02Async scrapers
03Deduplication
04Fit scoring
05Ollama tailoring
06Review queue

03 · Technical highlights

Implementation details

01

Aggregates structured APIs and browser-based sources through asynchronous scraping pipelines.

02

Deduplicates roles by external identifier and normalized title-company keys before persistence.

03

Uses a local Ollama model to generate factual cover letters in DOCX, PDF, and plain-text formats.

04

Stages LinkedIn Easy Apply workflows with Playwright while preserving a human approval boundary.

04 · What this demonstrates

Engineering signal

Practical AI integration without sending personal resume data to a hosted model provider.

Workflow automation spanning HTTP APIs, browser control, scheduling, persistence, and document generation.

Evidence-driven prioritization through configurable relevance scoring and review queues.

Continue exploring

View all projects