Data pipelinesJobs
aws-job-streamer
A live AWS pipeline that reads ~130 job boards and emails only the jobs worth applying to
~130job sources read every 4 hours, all in ~17 seconds
See the code on GitHub (opens in a new tab)
Proof you can check:Code on GitHub
Data pipelines
~130 job boards in, only the jobs worth it out
- ~130 sourcespublic job APIs, Workday boards, Adzuna, Remotive, USAJobs
- Can really be takenwork rights and location checked first
- New jobs onlyeach one scored once, ever: the AI bill stays small
- AI score 0–100one honest reason each (Claude Haiku on Amazon Bedrock)
- Ranked emailevery 4 hours, plus a daily heartbeat
Lubo Bali · lubobali.com
The problem
Job posts are spread over dozens of boards, many are stale, and the good ones fill fast. Reading them all by hand takes hours every week.
What I built
Every 4 hours a serverless pipeline pulls ~130 boards, drops jobs that can't be taken, scores each new one from 0 to 100 with an AI model and emails a ranked list. A daily "still alive" email and alarms mean it never fails silently.
Tech & why
Serverless AWS runs only on schedule, and Terraform rebuilds all of it from code. Each job is scored once, ever, so the AI bill stays small and predictable.
- AWS Lambda
- Terraform
- DynamoDB
- Amazon Bedrock
- Python
The result
Live and running: real matches arrive every 4 hours, 616 tests guard it, and a person still decides where to apply.
Similar projects
- Data pipelinesSQL Server → PostgreSQL migrationMoving a payments database to AWS, and proving the money still adds up83,892 report rows that silently stopped matching
- AI agentsJobRadar-AIAn AI agent that ranks thousands of job posts against one resume, and keeps the tracker up to date itself5,718 job posts ranked against one resume
- Data pipelinesTicketDesk-AIA support desk on Databricks Lakebase, built like a trading terminal180 tests, run against a real Postgres database
Have a problem like this? Get your free AI blueprint in about a minute.
