Inquire Now

Ensigncode is hiring a six-person GPU optimization team in Ahmedabad, each person owning a distinct part of the pipeline: kernels, profiling, inference systems, integration, numerical algorithms, and platform engineering. We assess capability and measured results over exact years of experience, and compensation has no fixed ceiling for the right candidate.

LocationAhmedabad · Full-time, permanent
ExperienceTypically 3–8 years, assessed on depth and results
CompensationNo salary bar for the right candidate
AvailabilityImmediate joiners or up to 15 days' notice preferred

Six specialists, one shared outcome

The goal isn't six people who all claim every skill. We hire for distinct strengths, then make the team effective through shared interfaces, code review, profiling evidence, and cross-training. Every role understands GPU fundamentals, but each owns a different part of the optimization pipeline.

Open roles

Current openings

Kernel owner

Senior CUDA Kernel Engineer

Kernel owner: designs and optimizes GPU compute kernels for GEMM, SpMM, and other compute-heavy workloads.

5-8 years preferred; strong candidates from 3+ years considered View role → Measurement owner

GPU Performance & Profiling Engineer

Measurement owner: builds benchmark suites and uses Nsight to diagnose GPU bottlenecks.

3-7 years View role → Runtime and workload owner

GPU Inference Systems Engineer

Runtime and workload owner: designs GPU inference execution flows, batching, scheduling, and throughput tuning.

4-8 years View role → API and developer experience owner

C++ / Python Integration Engineer

API and developer experience owner: builds stable C++ APIs and Python bindings for GPU functionality.

3-7 years View role → Algorithm and precision owner

Numerical & Compute Algorithms Engineer

Algorithm and precision owner: selects algorithms and validates numerical correctness across precision levels.

3-8 years View role → Deployment and device reliability owner

Embedded / GPU Platform Engineer

Deployment and device reliability owner: integrates GPU software with target systems and diagnoses platform-level issues.

3-7 years View role →

What every role needs

Shared hiring standards

These are team-wide expectations. We assess candidates on the skills relevant to their own specialty, not on being expert in every item below.

  • C++ fundamentals, debugging discipline, code quality, and collaborative engineering habits
  • GPU architecture basics: parallel execution, memory hierarchy, synchronization, and host/device data movement
  • A measurement-first mindset: establish a baseline, change one variable, benchmark, and report reproducible results
  • Numerical correctness and clear failure handling appropriate to the component they own
  • Comfort working on a team where kernel, profiling, runtime, integration, algorithm, and platform responsibilities are shared

What to expect

Our interview process

1. Screening

Relevant project ownership, GPU/C++ depth, notice period, and Ahmedabad work arrangement.

2. Specialty round

A role-specific practical discussion or code exercise, scoped to your specialty rather than all six.

3. Performance case

Diagnose a bottleneck from a small benchmark or profiling scenario and explain the trade-offs.

4. Team collaboration

How you work across interfaces, test assumptions, document results, and help adjacent specialists.

Ready to apply?

Don't see your exact role?

Apply anyway. If your background is a strong match for GPU, CUDA, or C++ performance engineering, we'll reach out even if it's not a perfect fit for one of the six roles above.

Apply for this role

Tell us about yourself and we'll get back to you about next steps.