Social Automation Platform Engineer (Full-Stack, AI/LLM)
Sandpiper Productions · Remote
About the role
Own the platform that turns our asset database into AI-generated, approval-gated, and auto-scheduled social campaigns across Meta and other channels. You’ll build the backend services, the UI module for composing/approving/scheduling, and the reporting pipes—so campaigns publish reliably with guardrails, observability, and traceability.
What you’ll do
- Compose → Approve → Schedule → Publish: Ship a UI module where teammates select assets from our DB, generate LLM variants, preview per platform, route for approval, and schedule posts.
- Integrations: Implement and maintain Meta Graph API (and peers) for OAuth, publishing, webhooks, pagination, rate-limit handling, retries, and error classification.
- Guardrails: Enforce brand/policy checks, blocked terms, size/format limits, and approval gates; log every action (create/edit/approve/publish) with auditability.
- Schedulers & workers: Build timezone-aware schedulers, queues, and backoff/retry strategies; surface live status via webhooks and alerts.
- Reporting & data: Attach campaign/UTM metadata automatically; ingest Insights/Analytics into the warehouse and power dashboards for performance and SLA monitoring.
- Reliability & security: Instrument traces/metrics/logs, set SLOs, create runbooks; implement RBAC, secrets management, and least-privilege access to the asset DB.
- Collaboration: Partner with Marketing/Brand to translate briefs into prompt templates and experiment plans; with Data to define schemas; with Legal on policy checks.
Outcomes you’ll own (KPIs)
- Automation rate: % of posts fully automated (target increases quarter over quarter).
- Reliability: Publish success rate, time-to-recover, and max retry thresholds within SLO.
- Throughput: Time from draft to scheduled publish; batched publishing without platform violations.
- Quality & lift: Engagement/CTR improvement of AI variants vs. human baseline (A/B).
- Data freshness: Insights ingestion latency SLA (e.g., ≤ X hours end-of-day).
- Compliance: Zero unauthorized publishes; 100% actions audit-logged.
Requirements
Minimum qualifications
- 4+ years building production web apps/services (Python or Node preferred), with queues/workers (e.g., Celery/RQ/SQS) and job schedulers.
- Hands-on with OAuth2 and Meta social platform API: publishing, webhooks, pagination, rate limits, error handling.
- LLM orchestration experience: prompt templates, content validation/moderation, safe outputs.
- Strong SQL; shipped ETL connectors; familiarity with a modern warehouse + BI (e.g., BigQuery/Redshift/Snowflake + Looker/Metabase).
- Front-end proficiency (React/TypeScript) to build a clean, role-aware workflow UI.
- Track record of shipping observability (metrics/logging/tracing) and writing runbooks/tests.
Nice to have (not required)
- TikTok/YouTube/X APIs; batch publishing at scale; media processing pipelines.
- Evals for generative content; brand safety/classifier experience.
- CI/CD, containers, IaC; SOC2-friendly audit logging.
- AB testing frameworks; feature flags; config/versioned prompts.
Benefits
Flexibility and competitive pay.