AI Inference & Compression Engineer: LLM & Video Codec Innovator

Beijing Foreign Enterprise Management Consultants Co.,Ltd.; Singapore; On-site; SGD 120,000 - 180,000; Full time

Job summary

Huawei is seeking an AI Inference & Compression Engineer in Singapore to drive LLM inference acceleration and advanced video compression techniques. You will work on KV cache optimization, model quantization, and memory bandwidth bottlenecks during autoregressive decoding, as well as develop AI-based media coding components and ensure smooth deployment within production frameworks.

The role covers evaluating visual quality and AI performance using metrics like PSNR, VMAF, perplexity, latency,

Qualifications

• Master’s or PhD in Computer Science, Electronic Engineering, Mathematics, or related fields (PhD preferred).
• Strong understanding of video coding fundamentals including prediction, transform coding, quantization, and entropy coding with hands-on experience in HEVC/AV1/H.266/VVC.
• Solid knowledge of Transformer architectures and attention mechanisms, and memory bandwidth constraints in AI inference.

Responsibilities

• LLM Inference Acceleration: research and develop compression algorithms to accelerate LLM serving.
• Classical Codec Development: design and implement advanced video compression algorithms to improve RD performance.
• AI-Based Media Coding: develop AI-based loop filters, optical flow, and rate control for video coding.
• Model Deployment & Fusion: optimize deep learning models for efficient inference and seamless integration of compression into deployment frameworks (e.g., vLLM).
• Performance & Quality Evaluation: assess PSNR, VMAF, perplexity, latency, and throughput for video and LLM systems.

Skills

Python

C/C++

PyTorch

TensorFlow

Transformers

LLM inference

Model optimization

Education

Master’s or PhD in CS/EE/Math

Tools

CUDA

vLLM

HEVC

AV1

Job description

Huawei is seeking an AI Inference & Compression Engineer in Singapore to drive LLM inference acceleration and advanced video compression techniques. You will work on KV cache optimization, model quantization, and memory bandwidth bottlenecks during autoregressive decoding, as well as develop AI-based media coding components and ensure smooth deployment within production frameworks.

The role covers evaluating visual quality and AI performance using metrics like PSNR, VMAF, perplexity, latency,