Reactivate Candidates Workflow: Ollama + Qdrant + Postgres (n8n)

Reactivate Candidates with AI: Ollama + Qdrant + Postgres in n8n

This n8n automation workflow takes a recruitment brief from an n8n form, expands it with Ollama, searches your historical candidate vectors in Qdrant, applies a code-based retention/consent compliance gate, then uses Ollama to rank matches and draft re-approach outreach—while logging every run to Postgres and returning an HTML shortlist.

What this workflow does

• Receives a role brief via an n8n form trigger.
• Builds a run ID, formats the brief, and computes retention cutoff dates based on your policy configuration.
• Expands the brief using Ollama into three descriptive search queries to improve recall across alternative titles and transferable backgrounds.
• Runs parallel semantic searches against a Qdrant collection (excluding candidates marked as consent withdrawn) and merges the results.
• Deduplicates candidates and enforces a compliance gate in code by calculating each candidate’s retention status from their last activity date—excluding ineligible records before the model sees them.
• Ranks and drafts outreach with Ollama, verifies returned IDs against the allowed set, and creates messages only for candidates marked contactable.
• Audits the run by writing records to Postgres and renders a consultant-facing HTML report with statuses and draft messages as the webhook response.

Use cases

• Re-activating candidates for a new role while respecting retention and consent rules.
• Reducing recruiter workload by generating ranked shortlists and outreach drafts from a brief.
• Reviewing an auditable AI candidate selection process via HTML output backed by Postgres logs.

Technical details

• Ollama: qwen2.5:7b (chat) and nomic-embed-text (embeddings).
• Qdrant: semantic search across a candidate vector collection.
• Postgres: stores an audit record per run.
• n8n nodes: code, merge, postgres, sticky note, form trigger, respond to webhook.