Description

You will engineer solutions focused on optimizing LLM performance at scale.

Responsibilities

• Analyze tracing logs from LLM inference and training runs to identify performance issues and inefficiencies.
• Develop tools and scripts to parse, visualize, and monitor LLM tracing data.
• Collaborate with ML and infra teams to recommend and implement performance optimizations.
• Investigate and resolve model latency and throughput issues related to runtime behavior.
• Contribute to best practices for performance tracing, benchmarking, and logging across model deployments.

Required Skills

• 5+ years of experience working with large-scale ML models, preferably LLMs (GPT, BERT).
• Proficiency in Python and ML Frameworks (PyTorch, TensorFlow).
• Familiarity with model tracing tools such as PyTorch Profiler, TensorBoard, or DeepSpeed.
• Experience with LLMs.
• Strong problem-solving skills for analyzing complex logs and metrics.
• Experience with MLOps.

Preferred Skills

• Experience with distributed training/inference and GPU performance optimization.
• Knowledge of systems profiling tools (e.g., NVIDIA Nsight, perf, Flamegraphs).

Key Skills

Education

Any Graduate