Event attendee personalisation platform

Title Full-stack developer — AI-powered “Why Attend” personalisation platform for B2B events Overview We run personalised “Why Attend” widgets for B2B trade shows and conferences. A visitor to an event website enters their job title (and optionally company), and receives an instantly generated report matching them to the most relevant exhibitors, sessions, and highlights for their role, along with a match/fit score. We have working front-end prototypes (single-file HTML, no build step) for two live events. We now want to move from static, manually-curated prototypes to a proper product: an automated content pipeline, a real matching engine, and a lightweight dashboard for event organisers. This is a fixed-scope build in phases. We’re looking for one developer (or small team) who can own the whole pipeline, not just the front end. What exists today • Two working widget prototypes (HTML/CSS/JS, single file, Netlify-deployable) • Persona sets and exhibitor/session data are currently hardcoded/placeholder • Visual design system (typography, colour, branding) is done per event What we need built Phase 1 — Content pipeline • A scraper/ingestion service that pulls exhibitor lists, session/agenda data, and speaker info from an event’s own website (and optionally uploaded documents — PDFs, spreadsheets) • Scheduled re-scraping to keep content current as the event approaches (configurable frequency) • Structured storage (exhibitors, sessions, personas, matches) in a proper database — not flat files Phase 2 — Matching engine • A scoring service that takes a visitor’s job title/company and returns ranked matching exhibitors and sessions • Should be more sophisticated than static keyword lookup — embeddings-based or LLM-assisted matching, tuned so scores are meaningfully differentiated across visitors, not clustered • API endpoint the front-end widget calls to generate a report on demand Phase 3 — Organiser dashboard • Lightweight admin panel for event organisers to: review/approve scraped content, see analytics (reports generated, click-through, top personas), and adjust branding • Basic auth (organiser accounts, not public) Phase 4 (stretch) — Distribution tools • Exportable HTML snippet for embedding reports in email campaigns • Simple public API or webhook so report data can be pulled into other tools Tech preferences • Backend: open to your recommendation, but should be something we can maintain post-handoff (Node/Python preferred over more niche stacks) • Frontend: keep the existing single-file HTML/CSS/JS widget pattern where possible; React only where it earns its complexity (e.g. the dashboard) • Hosting: Netlify-compatible front end; backend on a standard host we can access directly (no vendor lock-in you control) • No paid third-party SaaS dependencies without sign-off first Non-negotiables • All UI must meet WCAG AA contrast • UK English throughout (copy, docs, comments where user-facing) • We own the repository and all code outright; IP transfers on milestone payment (confirmed in contract terms) • Clean, commented handoff — we will be maintaining this ourselves after delivery • No personal data collection beyond job title/company at this stage (no SSO, no cookies, no email capture) — this is a deliberate product decision, not an oversight Ideal candidate • Demonstrable experience with web scraping/content extraction at production quality (handling site changes, rate limits, deduplication) • Comfortable with LLM API integration (embeddings, structured extraction, prompt-based scoring) • Can design a simple relational schema without over-engineering it • Has shipped at least one dashboard/admin panel from scratch • Good async communicator — we’ll work in milestones, not day-rate hours Engagement structure Fixed-price milestones, paid on delivery and review of each phase: 1. Content pipeline working end-to-end against our two existing events 2. Matching engine live behind an API, integrated with the existing widget 3. Organiser dashboard (MVP: content review + basic analytics) 4. (Optional extension) Distribution tools Please quote each phase separately in your proposal, along with your estimated timeline per phase. To apply, include • Examples of prior scraping/extraction or LLM-integration work (links or brief description — we understand some may be under NDA, a description of the problem and your approach is fine) • Your proposed stack for phases 1–2 • Any questions or scope concerns you’d flag before starting