JSE is an open-source, local-first job hunting tool. It scrapes job boards, scores each role against your resume with a tiered LLM assessment, and tracks everything on a Kanban pipeline. Your data stays on your machine in SQLite, and the LLM side runs against local endpoints (LM Studio, Ollama, llama.cpp), so nothing has to go to a cloud API. A RAG-based fragment memory pulls the relevant parts of your history into each application so tailoring doesn't start from a blank page. Stack is an Electron/React frontend with a Python backend and Selenium scrapers. Free to use, and I built it for my own job search first.

Local LLMOpen SourceJob SearchRAGProductivity

Looking forBeta users Feedback

About

I build hardware, create software, and I run the technology that keeps a business going. Most of my career has been spent across both.

Most recently I was IT and Engineering Manager at Flavorite, Australia's leading glasshouse food producer. I led a team of eight across applications, infrastructure, support and projects, ran a $5M annual operating and infrastructure budget, moved the business from MPLS to SD-WAN, lifted cyber controls against the Essential Eight, rolled out a labour management system to 1,200 staff, and merged systems through several acquisitions. Documented savings came to about $610k a year across procurement, licensing and carrier costs.

Before that I led engineering and product at Firetail, designing wireless down-hole monitoring hardware for oil and gas wells, around 1,000 m deep in hot, corrosive conditions. It ran on a $2M development budget with an 18-person field and agency team, and it turned a service business into a product business. I've also delivered a Salesforce CPQ build at EPSA, consulted independently on systems and growth, and worked at Bosch Power Tools on warranty systems, dealer training and digital strategy.

I'm completing a Bachelor of Mechatronics Engineering (Honours) at Deakin, studied part-time alongside full-time work. My thesis is an adaptive solar MPPT controller that learns online on a standard STM32H7, with no AI accelerator, tested against Perturb and Observe on a dual-channel GaN converter I built. Under partial shading it averaged 4.21 W where P&O fell to 0.84 W. The thesis is done, one subject is left, and I'm available for full-time work now. I also hold a Bachelor of Commerce.

I'm looking for my next role in clean energy, hardware or technology leadership, in eastern Melbourne or hybrid. If you're building something that has to work in the real world, I'd like to hear from you.