Description
Needs to hire 2 Freelancers Summary Developer to Build a Compliant Outbound Email System (Smartlead API + AI Personalization + Deliverability Monitoring) Code for the dashboard: https://www.dropbox.com/scl/fi/y4sryfm75rgeeowyckmjb/Biomeshield_Outbound_Starter.zip?rlkey=1b5nz46lm4woecqficyek08t2&dl=0 About the project We're a consumer wellness brand launching an influencer partnership program. We have a pre-qualified list of ~10,000 creators we want to reach out to individually about a product-gifting collaboration. This is a low-volume, high-personalization outreach program — think a few hundred thoughtful, individually tailored emails per day, not high-volume mass mailing. We want an engineer to build a small, well-architected system that runs this outreach cleanly and protects our sender reputation. Most of the "sending" is handled by an off-the-shelf platform (Smartlead) — your job is to build the integration layer, the AI personalization pipeline, and the monitoring/safety logic around it. Everything must follow email compliance law (CAN-SP1AM, and GDPR for any EU recipients) and email-provider deliverability best practices. Compliance and deliverability are non-negotiable requirements of this build, not afterthoughts. What you'll build Data ingestion (ETL). Parse our scored lead spreadsheet (.xlsx) into a PostgreSQL database. Honor our existing eligibility flags (some rows are pre-marked "exclude"). Make it idempotent — re-running never duplicates or double-sends. Email verification integration. Wire in a verification API (e.g. ZeroBounce or MillionVerifier) to scrub invalid/risky addresses before anything sends. Store the result per lead. AI personalization pipeline. An offline batch worker that calls the Claude API to generate a one-line, genuinely personalized opener per creator from public profile context we provide. Output is structured JSON (opener, niche, a quality score, and a flag for anything that should go to human review). Low-scoring outputs route to a review queue instead of auto-sending. Product claims come from a fixed, pre-approved copy block — the AI only writes the personalized intro line. Smartlead API integration. Push campaigns/sequences via Smartlead's API, map our sending mailboxes, and — critically — check our suppression list before every send. Webhook ingestion. A signature-verified, idempotent endpoint that receives Smartlead events (sent / open / click / reply / bounce / spam complaint / unsubscribe), updates lead state, and maintains a global suppression list. Deliverability monitoring. A scheduled poller that pulls Google Postmaster Tools data (and Yahoo's complaint feedback loop) per sending domain into the database — spam-complaint rate, domain reputation, and authentication (SPF/DKIM/DMARC) pass rates. Automatic safety "circuit breaker." Config-driven logic that automatically throttles or pauses sending when key health metrics cross safe thresholds (e.g. spam-complaint rate, bounce rate, or a drop in domain reputation), then follows a defined recovery/ramp-back procedure. All actions logged for audit. Dashboard + alerting. A lightweight health dashboard and Slack/email alerts + a daily digest (sends, bounce %, complaint %, reputation, replies, and any paused domains). Compliance & guardrails (please read — these are hard requirements) Every email includes a valid physical mailing address, honest sender/subject lines, and a working one-click unsubscribe (List-Unsubscribe header). Opt-outs and hard bounces are suppressed permanently and immediately — a suppressed address is never contacted again. We send from dedicated sending domains kept separate from our primary corporate domain — a standard deliverability practice to manage sender reputation. The system must never send from or authenticate against our primary domain. The system respects all provider rate limits and