Remote GPU Kernel & Inference Optimizer for Edge AI

Anyone Ai Inc.; Santa Fe Capital; A distancia; ARS 136.310.000 - 196.892.000; Jornada completa

Descripción de la vacante

Tether is seeking an AI Research Engineer (Kernel & Inference Optimization) to push the boundaries of model serving and inference on mobile and edge platforms. You will design high‑throughput, low‑latency pipelines and optimize memory footprint across diverse hardware.

Collaborating with cross‑functional teams, you will implement novel serving strategies, evaluate bottlenecks, and benchmark performance on real-world workloads.

Formación

• Deep expertise in designing and optimizing model serving pipelines and inference frameworks.
• Strong background in advanced model architectures and edge deployments.

Responsabilidades

• Design and deploy state-of-the-art model serving architectures that deliver high throughput and low latency while optimizing memory usage.
• Build, run, and monitor controlled inference tests in simulated and live production environments.
• Identify and prepare high-quality test datasets and simulation scenarios for real-world deployment on low-resource devices.
• Analyze computational efficiency and diagnose bottlenecks in the serving pipeline across processing and memory metrics.
• Collaborate with cross-functional teams to integrate optimized serving and inference frameworks into production pipelines for edge devices.

Conocimientos

MSL knowledge

GPU kernels

Model serving

Inference optimization

Edge devices

Diffusion models

Vision transformers

Pruning

Quantization

Flash attention

KV Cache

Speculative decoding

Educación

PhD in NLP/ML

Degree in Computer Science

Descripción del empleo

Tether is seeking an AI Research Engineer (Kernel & Inference Optimization) to push the boundaries of model serving and inference on mobile and edge platforms. You will design high‑throughput, low‑latency pipelines and optimize memory footprint across diverse hardware.

Collaborating with cross‑functional teams, you will implement novel serving strategies, evaluate bottlenecks, and benchmark performance on real-world workloads.