Member of Technical Staff, Inference & Serving

Inception; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

Inception is seeking engineers and scientists to design, optimize, and scale the diffusion LLM serving systems powering production inference. Your work will help make inference faster, more cost-efficient, and more reliable.

You will extend orchestration frameworks (Kubernetes, Ray, SLURM) for distributed inference, evaluation, and large-batch serving, and implement load balancing, autoscaling, and traffic routing for model endpoints.

Qualifications

• BS/MS/PhD in CS/Engineering or related field.
• Knowledge of ML serving frameworks (SGLang, vLLM, Triton Inference Server, TensorRT-LLM).
• Understanding of ML frameworks (PyTorch, TensorFlow) from a systems perspective.
• Familiarity with HPC and GPU programming (CUDA).
• Experience with containers (Docker), orchestration (Kubernetes), and CI/CD pipelines.
• Background in performance optimization and profiling of ML systems.

Responsibilities

• Build and optimize high-performance model serving systems for low-latency inference of diffusion LLMs.
• Extend orchestration frameworks (Kubernetes, Ray, SLURM) for distributed inference, evaluation, and large-batch serving.
• Implement and manage load balancing, autoscaling, and traffic routing for model endpoints.
• Build systems for model versioning, canary deployments, and zero-downtime rollouts.
• Develop monitoring, alerting, and observability tooling to ensure SLA compliance and rapid incident response.
• Collaborate with ML researchers to translate model advances into production-ready serving improvements.

Skills

Distributed inference

Performance optimization

Profiling ML systems

Collaboration with ML researchers

Education

BS/MS/PhD in CS/Engineering

Tools

Docker

Kubernetes

CI/CD

CUDA

PyTorch

TensorFlow

Kubeflow

Airflow

SGLang

vLLM

Triton Inference Server

TensorRT-LLM

Job description

Overview

We\'re looking for engineers and scientists to design, optimize, and scale the systems that power our diffusion LLMs in production. Your work will make inference faster, more cost-effective, and more reliable.

Key Responsibilities

• Build and optimize high-performance model serving systems for low-latency inference of diffusion LLMs.
• Extend orchestration frameworks (Kubernetes, Ray, SLURM) for distributed inference, evaluation, and large-batch serving.
• Implement and manage load balancing, autoscaling, and traffic routing for model endpoints.
• Build systems for model versioning, canary deployments, and zero-downtime rollouts.
• Develop monitoring, alerting, and observability tooling to ensure SLA compliance and rapid incident response.
• Collaborate with ML researchers to translate model advances (new architectures, quantization techniques, batching strategies) into production-ready serving improvements.

Qualifications

• BS/MS/PhD in Computer Science, Engineering, or a related field (or equivalent experience).
• Knowledge of ML serving frameworks (SGLang, vLLM, Triton Inference Server, TensorRT-LLM).
• Understanding of ML frameworks (PyTorch, TensorFlow) from a systems perspective.
• Familiarity with high-performance computing and GPU programming (CUDA).
• Experience with containerization (Docker), orchestration (Kubernetes), and CI/CD pipelines.
• Background in performance optimization and profiling of ML systems.

Preferred Skills

• Experience building and maintaining large-scale language models with tens of billions of parameters or more.
• Experience with distributed systems and cloud computing platforms (AWS/GCP/Azure).
• Experience with ML workflow orchestration tools (Kubeflow, Airflow).
• Experience with model optimization techniques (quantization, distillation, speculative decoding, continuous batching).
• Knowledge of ML-specific infrastructure challenges (checkpointing, resource scheduling, etc.).