Senior AI Scientist: Deep Learning & GPU-Optimized Production

Avrioc Technologies; Abu Dhabi; On-site; AED 240,000 - 320,000; Full time

Job summary

Avrioc Technologies in Abu Dhabi is seeking a hands-on Data Scientist to develop, train, optimize, and deploy AI models in production environments with expertise in deep learning, NLP, computer vision, and speech processing.

You will work on multi-modal large-scale use cases, optimize GPU-based training and inference, and take models from ideation to scalable deployment in collaboration with engineering and product teams.

Qualifications

• 4+ years of professional experience in AI:Model Training / Inference Optimization / Distributed Systems / Systems Programming / GPU Programming.
• Hands-on experience with Deep Learning model training.
• Strong programming skills in C/C++, Rust, Zig-lang.
• Strong experience with libtorch / PyTorch ecosystem.
• Hands-on experience with open-source inference engines (vLLM, SGLang, Llama.cpp).
• Hands-on experience with CUDA, RoCM and their template libraries.
• Strong Linux and shell scripting experience.

Responsibilities

• Develop, train, validate, and optimize frontier AI models.
• Work on Multi-Modal large scale use cases.
• Inference and training recipe optimizations for GPU-based environments.
• Take models from ideation, training and inference scaling through production deployment.
• Collaborate with engineering and product teams to deliver scalable AI solutions.

Skills

Deep Learning

NLP

Computer Vision

Speech Processing

Distributed Systems

GPU Programming

C/C++

Rust

Zig-lang

PyTorch

libtorch

CUDA

Linux

Shell scripting

Docker

Kubernetes

MLFlow

ONNX

Llama.cpp

vLLM

SGLang

Tools

Llama.cpp

vLLM

SGLang

Job description

Avrioc Technologies in Abu Dhabi is seeking a hands-on Data Scientist to develop, train, optimize, and deploy AI models in production environments with expertise in deep learning, NLP, computer vision, and speech processing.

You will work on multi-modal large-scale use cases, optimize GPU-based training and inference, and take models from ideation to scalable deployment in collaboration with engineering and product teams.