Senior AI Infrastructure Engineer — Agentic Platform & LLM

Snorkel AI; San Francisco (CA); On-site; USD 220,000 - 260,000; Full time

Job summary

Snorkel AI is seeking an experienced Platform/Infra engineer to help build the agentic stack and the AI infrastructure powering data-centric AI workflows. You will design an orchestration layer, memory and knowledge graphs, and governance for enterprise-scale agents and tools.

You’ll also develop the cost- and performance-focused LLM platform, improve data access layers across Snowflake, S3, and RDS, and ensure reliability with observability and RBAC governance across the platform.

Qualifications

• 4+ years building platform infrastructure, data infrastructure, or backend systems with significant data components.

Responsibilities

• Design and build the opinionated agentic stack that teams will use to scaffold agent workflows.
• Build the LLM cost and efficiency platform to control token spend and optimize costs.
• Develop the foundational data access layer and SDKs for multi-source data reading/writing.
• Design and implement event-driven data flows with brokers, CDC, schema registry, and routing.
• Build governance, lineage, and audit infrastructure for data and agents.

Skills

Platform infrastructure

Python

LLM in production

AI-assisted development tools

SQL

Snowflake

Redshift

Postgres

AWS

Kubernetes

Data orchestration (Prefect/Airflow/Ds

OpenTelemetry

RBAC/PII governance

Tools

Snowflake

Redshift

Postgres

Prefect

Airflow

Dagster

dbt

S3

RDS

EKS

EventBridge

Job description

Snorkel AI is seeking an experienced Platform/Infra engineer to help build the agentic stack and the AI infrastructure powering data-centric AI workflows. You will design an orchestration layer, memory and knowledge graphs, and governance for enterprise-scale agents and tools.

You’ll also develop the cost- and performance-focused LLM platform, improve data access layers across Snowflake, S3, and RDS, and ensure reliability with observability and RBAC governance across the platform.