Senior AI Inference Architect

Dizzaract; United Arab Emirates; On-site; AED 300,000 - 600,000; Full time

Benefits offered by this job

Real technical ownership

Competitive salary with performance‑dr

Health insurance

24 days annual leave

Modern office in Yas Creative Hub

Continuous learning

Job summary

Dizzaract’s FAR Labs seeks a Lead Inference Engineer to own the AI serving stack and measurement. You’ll drive performance across inference serving, GPU utilization, and heterogeneous hardware, setting standards and guiding engineers in a senior, hands-on capacity.

This role combines architecture, benchmarking, and practical implementation to push fast, efficient, scalable AI serving in production environments.

Qualifications

• Hands-on experience building and operating production LLM inference serving systems.
• Deep understanding of vLLM, SGLang, TensorRT-LLM or similar stacks.
• Expertise in paged attention, continuous batching, KV-cache management.
• Strong GPU performance engineering with CUDA or Triton, FP8/FP4.
• Experience serving MoE models with expert parallelism and load balancing.
• Model quantization and measurement against full-precision references.
• Robust benchmarking: load generation, latency percentiles, reproducibility.
• English level: Advanced (C1).

Responsibilities

• Own and improve the FAR Labs inference serving stack and caching.
• Improve throughput, latency, memory efficiency and cost while upholding model quality.
• Identify GPU bottlenecks and optimize using CUDA, Triton, and profiling.
• Drive efficient MoE serving with expert parallelism and all-to-all comms.
• Implement and evaluate quantization and other model optimisations.
• Build rigorous benchmarking and measurement methodologies.
• Set technical roadmap and engineering standards for production readiness.
• Lead and mentor engineers across serving, metrics, and testing.

Skills

LLM inference

Performance engineering

GPU‑Native

Distributed systems

Measurement‑driven

Technical leadership

AI infrastructure

English (Advanced)

Tools

vLLM

SGLang

TensorRT-LLM

CUDA

Triton

Job description

Dizzaract’s FAR Labs seeks a Lead Inference Engineer to own the AI serving stack and measurement. You’ll drive performance across inference serving, GPU utilization, and heterogeneous hardware, setting standards and guiding engineers in a senior, hands-on capacity.

This role combines architecture, benchmarking, and practical implementation to push fast, efficient, scalable AI serving in production environments.