AI SDK & Platform Engineer: Edge AI Runtime Architect

I Machines, Inc.; United States; On-site; USD 150,000 - 230,000; Full time

Benefits offered by this job

Health, dental, vision

Retirement savings plans

Paid time off

Flexible Schedule

Job summary

I Machines, Inc. is seeking an experienced AI SDK Engineer to build a full software platform that lets customers move models from major frameworks to our AI compute engines—from edge to cloud.

You will design a high-performance runtime, craft optimized kernels, and enable end-to-end deployment across devices and platforms. The role requires 5+ years in AI infrastructure or compiler/runtime work, strong C/C++ and Python skills, and hands-on experience with PyTorch, TensorFlow, ONNX, and

Qualifications

• BS/MS/PhD in Computer Science, Engineering, or equivalent.
• 5+ years in systems software, AI infrastructure, compiler/runtime, or embedded AI.
• Excellent C/C++ and Python proficiency.
• Experience deploying AI models on specialized hardware.

Responsibilities

• Build end-to-end AI model deployment pipelines across major frameworks for iMachine hardware.
• Design and maintain a high-performance AI runtime with multi-accelerator support.
• Create optimized kernels for core AI ops (GEMM, attention, conv, KV-cache, quantization).
• Optimize LLM/VLM inference performance including KV-cache and low-precision formats.

Skills

C/C++

Python

Systems software

AI inference

Multi-accelerator

Education

BS/MS/PhD in CS/CE/EE

Tools

PyTorch

TensorFlow

ONNX

LLVM/MLIR

Job description

I Machines, Inc. is seeking an experienced AI SDK Engineer to build a full software platform that lets customers move models from major frameworks to our AI compute engines—from edge to cloud.

You will design a high-performance runtime, craft optimized kernels, and enable end-to-end deployment across devices and platforms. The role requires 5+ years in AI infrastructure or compiler/runtime work, strong C/C++ and Python skills, and hands-on experience with PyTorch, TensorFlow, ONNX, and