Forward-Deployed Inference Engineer (Remote | Equity)

Fdetimes; San Francisco (CA); Remote; USD 270,000 - 300,000; Full time

Benefits offered by this job

Startup equity

Health insurance

Flexible remote work

Job summary

Together AI, the AI Native Cloud, is seeking a Forward Deployed Engineer focused on inference to partner with strategic customers and production AI teams. You will optimize inference engines, fine-tune pipelines, and guide deployments to achieve targeted throughput and latency.

You will work with SAs as a domain expert, influence customer success, and contribute to platform improvements while supporting high-value POCs and scalable deployments.

Qualifications

• 5+ years in a technical role with focus on system optimization.
• Hands-on experience with inference engines (vLLM, TensorRT-LLM, SGLang).
• Deep knowledge of KV cache tuning, speculative decoding, tensor parallelism, and quantization.
• Broad knowledge of state-of-the-art open-source models and model selection.
• Strong Python skills and production-level experience.

Responsibilities

• Optimize inference engines for given hardware and workloads.
• Develop configuration updates to win POCs and benchmarks; tune caches and decoding strategies.
• Act as primary technical contact for strategic accounts; align deployments with milestones.
• Onboard strategic customers with correct inference configurations from day one.
• Help shape product roadmap by surfacing field insights and driving feature adoption.

Skills

Python

Project management

Communication skills

Tools

vLLM

TensorRT-LLM

SGLang

Job description

Together AI, the AI Native Cloud, is seeking a Forward Deployed Engineer focused on inference to partner with strategic customers and production AI teams. You will optimize inference engines, fine-tune pipelines, and guide deployments to achieve targeted throughput and latency.

You will work with SAs as a domain expert, influence customer success, and contribute to platform improvements while supporting high-value POCs and scalable deployments.