Software Engineer, Networking (Agent Infra)

present of AI; San Francisco (CA); On-site; USD 140,000 - 210,000; Full time

Job summary

Baseten is seeking a Software Engineer, Networking (Agent Infra) to design and operate a user-space data plane that scales across thousands of sandboxes. You will build multi-tenant private networking, overlays, routing, and secure ingress/egress, using Rust, Go, or C/C++.

You’ll work close to platform infra and optimize for latency, throughput, and resilience in AI workloads. You should have hands-on experience with VPP/DPDK, TCP/IP, and cloud networking concepts, and be ready to ship

Qualifications

• Production-grade systems programming in Rust, Go, or C/C++.
• Experience building cloud networking as provider, not just consumer.
• Hands-on with VPP/DPDK or virtual network functions.
• Strong TCP/IP and routing knowledge.
• Familiarity with overlay networks and multi-tenant isolation.

Responsibilities

• Build and run a user-space packet processing data plane for many sandboxes.
• Implement multi-tenant private networking, overlays, routing, and secure ingress/egress.
• Debug latency and packet loss under real load and optimize performance.
• Collaborate with infrastructure teams to support AI product runtime.

Skills

Rust

Go

C/C++

Networking

TCP/IP

Linux

Tools

VPP

DPDK

VXLAN

GENEVE

Linux kernel networking tools

Job description

Software Engineer, Networking (Agent Infra)

This role asks for an engineer who can build and run a user space packet processing data plane for large numbers of isolated sandboxes, including multi tenant private networking, overlays, routing, and secure ingress and egress. A strong candidate has shipped production systems programming in Rust, Go, or C/C++ and has built cloud networking as the provider rather than only using one. Hands on work with DPDK, VPP, or virtualized network functions, plus solid TCP/IP and routing knowledge, matters more than configuration experience.

How it fits the AI era

AI agents now run in huge numbers of short lived sandboxes that constantly start, stop, and talk to models, tools, and each other, and ordinary cloud networking was not built for that pattern. Much of the routine configuration and monitoring can be automated, but designing a fast, isolated, multi tenant data plane and debugging latency and packet loss under real load stays human work. The role sits close to the infrastructure that makes AI products run, so the engineering judgment about scale and security is the durable part.

How it builds AI experience

Working here means gaining direct experience with the networking layer that agent platforms depend on, from microVM connectivity to programmable network APIs. A candidate can build this by studying user space packet processing, overlay protocols, and proxy and load balancing internals, and by practicing packet capture and profiling on real traffic. Familiarity with VPP, GENEVE or VXLAN, and Linux kernel networking tools would strengthen a case for this team.

Worth showing

• VPP or DPDK
• Multi tenant overlay networking
• Rust Go or C/C++
• Packet capture and profiling

This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.

About Baseten

High-performance cloud inference platform for deploying, serving, and scaling open-source and custom artificial intelligence models.

Tools mentioned: Go, Rust, C++, Spark, Linux, machine learning