The role focuses on optimizing inference engines for local and edge environments, ensuring models run efficiently on hardware owned by users. The team is dedicated to building agentic AI that balances local and cloud intelligence while maintaining data privacy and cost-effectiveness. Candidates will work with technologies like llama.cpp and vLLM, tuning performance for latency, throughput, and memory. This position is critical for developing low-cost AI solutions that are both powerful and trustworthy.

• BS/MS in Computer Science, Electrical Engineering, Mathematics, or a related STEM field
• 5+ years of software development experience
• Strong proficiency in C++ and/or Python, with the ability to read systems-level code
• Understanding of how LLM inference works, including attention, KV cache, and decoding
• Experience in profiling and optimizing real performance problems on CPU or GPU
• Expertise in Linux, build systems, and low-level debugging

• Profile and optimize local inference for latency, throughput, and memory on edge hardware
• Tune KV cache, continuous batching, and scheduling for interactive agent workloads
• Drive quantization strategy and validate quality impact with the Post-Training team
• Reduce CPU overhead and improve engine startup, model load, and lifecycle management
• Benchmark across hardware tiers and publish performance comparisons
• Contribute upstream fixes and patches to open-source engines where beneficial

• Intel offers a comprehensive benefits package including competitive pay, stock programs, healthcare coverage, retirement plans, paid time off, parental leave, and programs supporting employee wellbeing and professional development.