About the job
As a Staff Engineer, Agentic at Inflection, you will own the platforms, systems, and services that bring our conversational AI to life at scale. You’ll collaborate across research, product, and infrastructure teams to enable rapid iteration, high reliability, and secure delivery of novel AI features to millions of users. Your work will directly impact both the pace of product development and the stability of our production systems.
Responsibilities
Design and build backend systems that power agentic workflows in production: orchestration, memory, retrieval, and tool integrations.
Operate high-availability inference and conversation pipelines, optimizing for latency, cost, and quality.
Build internal platform pieces service templates, CI/CD, observability, evaluation harnesses, and rollout tooling that compound engineering velocity.
Partner with applied researchers and frontend engineers to design and iterate on user-facing features
Drive features from prototype to production with measurable user impact, ensuring reliability and performance at scale
Qualifications
Minimum
10-15 years of backend engineering experience shipping production systems in cloud environments.
Demonstrated experience building, shipping, and operating agentic AI systems in production environments with real user impact not just prototypes.
Strong production experience with Python, Postgres, Redis, and at least one major cloud (AWS, GCP, or Azure), including containers and Kubernetes.
Practical experience with LLM application patterns: RAG, tool use, evaluation loops, and the latency/cost/quality tradeoffs that come with them.
Strong sense of ownership, with the ability to independently drive systems to production and operate them reliably
Preferred
Experience with agent orchestration frameworks (LangGraph, LlamaIndex, OpenAI/Anthropic Agent SDKs, or equivalent in-house systems) and evaluation tooling (LangSmith, Braintrust, or similar).
Background in streaming, queueing, or vector data infrastructure at scale.
Past on-call ownership of customer-facing services with strict SLOs.
Contributions to open-source agent or LLM tooling.