Description
Summary We're a small B2B agency and our team is drowning in manual lead research. Every week someone spends hours pulling company and contact data from directories and LinkedIn, copying it into spreadsheets, cleaning it up, and figuring out which leads are worth chasing. It's slow, it's inconsistent, and good leads slip through the cracks. We want to automate this end to end. The idea is a sourcing pipeline that takes a source (a directory, a job board, or an API), pulls the raw list, cleans and standardises the data into a consistent format, enriches it with company details, scores each lead against our criteria, and drops the finished list into our CRM ready to use. We've started building bits of this in n8n but it's fragile, it breaks the moment a source returns messy or incomplete data, and nobody on our side can fix it. We need someone who actually knows how to build these properly, with real error handling so it doesn't fall over in production. You'll work with n8n, the Claude or OpenAI API for the enrichment and scoring logic, and some custom Node where the no-code tools hit their limits. A simple dashboard to review results would be a bonus. What we care about: reliability (it has to work every run, not just the demo), clean structured output, and clear communication so we understand what you built. This is a small first project but we have a steady stream of automation work if it goes well.