About new AI Startup staring in few months:

This startup is going to work on Offload observability and debugging to make AI workload reliable and work consistantly across any hardware that tells operators what is actually running in DPU/SmartNIC hardware versus software and it turns hardware offload from a black box into something you can monitor, alert on, and debug.

Principal Engineer - Systems and Data Path

Job Description

Experience in some of the below technologies with C, C++, Rust, or similar low-level languages for bare-metal performance engineering in hyperscale or cloud-provider environments

Experience in low-level network programming.

Network and Virtualization background in Linux network stack internals, socket layers, and packet processing, understanding of the VMs, specifially disks storage and networking.

Strong understanding of Linux operating system internals.

Experience with eBPF or advanced Linux kernel networking.

Familiarity with embedded systems, SmartNICs, DPUs, programmable hardware accelerators, or hardware-adjacent software development.

Deep background in SR-IOV, NFV (Network Functions Virtualization), and PCIe device pass-through.

Experience investigating or implementing hardware interconnect acceleration ASIC pipeline architectures and networking hardware (such as NVLink GPU-to-GPU performance optimizations)

Hands-on experience with Linux OS internals, Debian build/packaging, and kernel networking stack.

Experience with DPDK, P4, BCM SDK, or other hardware programming interfaces

Experience with Linux networking internals of routing, netfilter/iptables/nftables, conntrack in production environments.

Hands on experience with network I/O drivers and L2-L7 networking stack consisting of data plane software such as DPDK, VPP, etc.

Experience supporting AI/ML, GPU, or high-performance computing networking environments.

Architect and implement high performance, highly available system software components – DPDK-based networking stack, hardware abstraction layers, and custom hardware acceleration stacks.

Experience in SDK development for Ethernet switch silicon

Responsibilities

Deep expertise in high-performance networking, including Ethernet, RDMA, RoCE, PCIe, DMA, and modern NIC architectures.

Expertise in high-performance networking frameworks and techniques such as DPDK, eBPF/XDP, AF_XDP, io_uring, kernel bypass, zero-copy I/O, and RDMA/RoCEv2

Experience with networking software stack components for AI applications such as RoCE, libFabric, OpenMPI, NCCL, OneCCL, etc.

Leverage Linux OS expertise to manage memory, scheduling, I/O virtualization, kernel drivers and security features of the DPUs. 

Experience with RIB, FIB, HAL, and high-availability (HA) networking solutions.

Ability to debug complex software issues across operating system, driver, firmware, and hardware boundaries

Experience collaborating closely with hardware and ASIC design teams, contributing to hardware specifications, co-designing hardware-software interfaces, and influencing silicon feature sets based on software requirements

Direct experience with DPU SoC/SmartNIC Platforms

Experience in hyperscale or cloud-provider environments

Collaborate with DPU vendors to implement software to efficiently use hardware resources on the DPU.

Ability to debug complex software issues across operating system, driver, firmware, and hardware boundaries.

Experience with large-scale cloud providers or cloud service architectures.

Qualifications

B.Tech or B.E in Electrical or Electronics or Computer Engineering or Computer Science with hands-on system development/testing experience.

15+ years of networking, infrastructure, or cloud engineering experience.

Contact

intersted candidates can send thier resume to info@epistemesoft.com