Senior/Staff Backend Engineer (Node.js/Python) - AI Learning Infrastructure

Senior or Staff Backend Engineer (Node.js/Python) | Self-Learning AI Infrastructure | Miami, FL

About the Company

We are a confidential, high-velocity AI startup in Miami building the learning and memory infrastructure for an Agentic platform serving the global SMB technology market and the broader IT service ecosystem. The product centers on autonomy, real-time reasoning, and continuous improvement. You will architect systems that help AI learn every night and perform reliably during the day.

About the Role

In roughly 18 months you will ship systems most engineers never touch, progressing from senior developer to architect of autonomous learning systems. The company has approximately $6.5M in pre-seed funding and is focused on making AI economically sustainable as automation accelerates across industries.

Responsibilities

• You will own the autonomy stack by designing, building, and operating data pipelines, API integrations, and key third-party connections that power learning and inference.
• You will implement the learning loop through metacognition pipelines that detect mistakes, capture feedback, and drive memory consolidation so the system gets smarter each night.
• Your mandate includes delivering infrastructure that scales from hundreds to hundreds of thousands of users with high data integrity, idempotency, and backpressure controls.
• You will help evolve the knowledge layer from vector search to causal and graph-based reasoning so the platform can shift from costly general models to efficient specialist intelligence.
• In your first 90 days, you will lift engineering throughput through better tooling, CI/CD, observability, safe rollout mechanisms, and lead the reliability posture with deployment automation, SLOs, error budgets, and thoughtful on-call practices.

Qualifications

• 5+ years in backend or data engineering with production Python and solid Node.js or TypeScript for real-time services.
• Experience building event-driven systems with Pub/Sub or Kafka that meet millisecond-level latency targets and handle failures gracefully.
• Strong in GCP (BigQuery, Firestore, Pub/Sub, Cloud Functions or Cloud Run) or equivalent depth on AWS or Azure with a convincing migration story.
• Ability to own problems end to end across data, APIs, infrastructure, and reliability, think in systems, communicate crisply, and prefer shipping version one and iterating quickly.

Required Skills

• Experience with Python for data transformation, orchestration, and ML pipeline integration (including asyncio, Pydantic, and FastAPI where appropriate).
• Node.js with TypeScript for high-concurrency, event-driven services and real-time APIs (Express or Fastify, NestJS familiarity is a plus).
• GCP (similar tools in AWS or Azure works too!).
• Streaming relies on Google Cloud Pub/Sub or Kafka with DLQs, replay capabilities, and schema evolution using Protobuf or Avro.
• BigQuery serves as the analytical backbone with partitioning, clustering, materialized views, and cost controls.
• Firestore supports operational documents and real-time sync with careful denormalization.
• Redis for caching and queues, PostgreSQL for relational workloads, and ClickHouse or Elastic for time-series and search when needed.
• Processing and orchestration include Cloud Functions or Cloud Run for serverless transforms, Dataflow or Apache Beam for streaming and batch, and Cloud Composer or Dagster for DAGs.
• Data quality is enforced through contracts and validators, with idempotent processors and exactly-once semantics where feasible.
• The platform runs on Docker and Kubernetes (GKE), is provisioned with Terraform, and shipped via GitHub Actions with canaries and progressive delivery.
• Observability uses OpenTelemetry, Prometheus, Grafana, Cloud Logging and Monitoring, and structured tracing.
• Security patterns include OAuth2 or OIDC, scoped tokens, secrets management, audit trails, and least-privilege IAM.

Preferred Skills

• Background in knowledge graphs or graph databases, feature stores, or large-scale Beam or Dataflow.
• Experience with real-time sync such as WebSockets or Server-Sent Events at high concurrency.
• Familiarity with retrieval pipelines, vector stores, guardrails, prompt and output validation, and feedback capture for nightly improvement as we expand the agentic layer and adopt graph-based reasoning for durable memory.

This is a hybrid role in Miami, Florida. Relocation assistance is provided. There is no remote option. You will work directly with technical leadership and product decision makers in a hands-on environment.