Roles & responsibilities :

- Deploy and manage ML/LLM models (<20B parameters) in production environments.

- Design and execute model load testing and performance benchmarking (latency, throughput, memory, cost).

- Build and optimize multi-node, multi-GPU training pipelines.

- Configure and tune distributed training frameworks (data parallelism, model parallelism, pipeline parallelism).

- Optimize GPU utilization, memory footprint, and inference costs.

- Set up CI/CD pipelines for model deployment and retraining.

- Troubleshoot GPU, networking, and performance bottlenecks.

- Work across cloud platforms to ensure portability and vendor-agnostic deployments.

Skill Sets :

- Strong experience with GPU workloads (NVIDIA GPUs, CUDA concepts).

- Proven expertise in model deployment on AWS, Azure, and GCP.

- Hands-on experience deploying models up to 20B parameters.

- Experience with distributed training (multi-node, multi-GPU setups).

- Deep understanding of load testing, stress testing, and benchmarking ML systems.