AI Engineer - Cortex Code Quality

Snowflake Snowflake · Data AI · CA-Menlo Park, United States · Engineering

AI Engineer focused on building and deploying coding agents for data applications within Snowflake's Cortex platform. The role involves owning the end-to-end AI engineering lifecycle, including agent design, prompt/tool engineering, evaluation, deployment, and optimization, with a strong emphasis on enterprise-grade features like function calling, tool schemas, guardrails, and agent teams. The position also requires designing and implementing evaluation pipelines and metrics to improve agent performance.

What you'd actually do

  1. Own features end-to-end for Snowflake Cortex Code products. Build agentic workflows, coding harnesses, evaluation pipelines.
  2. Build enterprise-grade context engineering: function calling, tool schemas, guardrails, agent teams, and verification/repair.
  3. Design evals and hillclimb : create golden sets, create rubrics and metrics, analyze errors, run experiments to hill climb on the metrics.
  4. Partner with product and infra: translate customer problems into products and experiments. Collaborate with infrastructure teams to productionize improvements.
  5. Work with an elite team of engineers towards building great products

Skills

Required

  • Python
  • Typescript
  • Go
  • agentic workflows
  • evaluation pipelines
  • function calling
  • tool schemas
  • guardrails
  • agent teams
  • verification/repair
  • golden sets
  • rubrics
  • metrics
  • error analysis
  • experimentation

Nice to have

  • dbt
  • airflow
  • data modeling
  • data analysis
  • retrieval systems
  • semantic layers
  • agentic coding tools
  • LLM observability
  • safety / guardrails

What the JD emphasized

  • shipping AI features in production
  • built and owned complex systems
  • production reliability

Other signals

  • building the future of coding agents for working with data
  • own the full AI engineering lifecycle: design, prompt/tool engineering, evals, deployment, measurement, and optimization
  • build enterprise-grade context engineering: function calling, tool schemas, guardrails, agent teams, and verification/repair
  • design evals and hillclimb