Senior Inference Engineer, AGI

What you'll do

• Co-design model architectures and optimize inference paths, GPU kernels, and performance for multimodal models.
• Build and operate real-time streaming serving systems, including scheduling, batching, multi-GPU inference, and performance benchmarking.
• Develop offline inference and rollout infrastructure for reinforcement learning and evaluation, ensuring train/serve consistency.

What they ask for

• 3+ years of building machine learning models for business application experience; 2+ years of hands-on experience optimizing neural-model inference.
• PhD, or Master's degree and 6+ years of applied research experience.
• Experience programming in Java, C++, Python or related language; experience with neural deep learning methods and machine learning.
• Strong understanding of deep learning architectures and speech/audio or multimodal applications, GPU performance optimization, and production latency-constrained real-time inference under concurrent load.

Nice to have

• Production LLM/multimodal serving internals, custom GPU kernels, or real-time streaming AI systems.
• Offline inference or reinforcement-learning infrastructure, distributed training, speech/audio generative models, or shipping research models to production.

Pay

USD 192,200.00 - 260,000.00 annually (base salary); USD 167,100.00 - 226,100.00 annually (base salary)

Tools

Java, C++, Python, vLLM, PyTorch, Nsight Compute/Systems, INT8/FP8/BF16, NCCL

AI focus

Real-time multimodal conversational AI inference, including speech and audio models, inference optimization, reinforcement-learning infrastructure, and evaluation.