Location: Bangalore [Adaptable]

Experience: 12 to 18 Years Exp

Type: Full time

About Us

We’re building the next generation of AI SmartNICs — intelligent network accelerators designed to power distributed AI training, data center virtualization, and hyperscale cloud workloads.

Our mission: push networking performance to the limits — from microsecond-level latency to terabit-scale throughput — while giving developers full programmability across compute, network, and storage fabrics. If you love working close to the metal and want to shape the future of data-driven infrastructure, you’ll fit right in.

What You’ll Do
Own and build NIC and RDMA kernel drivers for our SmartNIC platform (Linux).
Design and implement high-performance Tx/Rx and RDMA data paths — QPs, CQs, doorbells, DMA, and WQEs.
Bring up new silicon: work directly with hardware and firmware teams to validate PCIe, queue interfaces, and offload engines.
Add networking offloads (TSO, checksum, RSS, VLAN, tunneling: VXLAN/Geneve/GRE).
Integrate RoCEv2 and user-space RDMA (libibverbs/rdma-core) for AI and storage workloads.
Build tools for performance measurement, kernel tracing, and packet analytics.
Collaborate across firmware, DPDK,



and P4/eBPF pipeline teams to build a unified data path.
Push upstream when possible — we’re believers in open standards and open source.

What You’ll Bring
12+ years of Linux kernel or driver development experience (preferably network or RDMA subsystems).
Robust in C programming, concurrency, and debugging at the kernel level.
Solid understanding of:

NIC architecture — Tx/Rx rings, CQEs, doorbells, DMA engines. RDMA fundamentals — QPs, PDs, MRs, WQEs, CQ handling.

PCIe, MMIO, and DMA interactions.
Experience with rdma-core, DPDK, or low-latency networking stacks.
Familiarity with SR-IOV, VFIO, and virtio.
Passion for performance tuning and squeezing out every nanosecond.

Bonus Points
You’ve worked with SmartNICs or DPUs (NVIDIA BlueField, Pensando, Broadcom Stingray, Marvell Octeon, etc.).
You understand how AI/ML training clusters use RDMA for scale-out.
Experience with eBPF/XDP, P4, or programmable data planes.
You’ve contributed to Linux netdev or RDMA subsystems.
You enjoy profiling, tracing, and reverse-engineering hardware behavior.

📌 Senior Software Engineer – Nic & Rdma Driver Next Gen Ai Smartnic Bengaluru
🏢 Ideaon
📍 Bengaluru