Summary

Join an ai inference engineering team building and operating the inference engine behind perplexity queries. You will support transformer retrieval, text-generation, and multimodal models, migrate gpu kernels to nvidia cute dsl, and develop a rust-native inference server for low-latency, cost-aware serving.

Responsibilities

• Support new model architectures in inference
• Manage weight loading and request scheduling
• Implement kv-cache management for inference
• Port cuda kernels to nvidia cute dsl
• Develop a rust-based inference serving runtime
• Profile and fix performance bottlenecks
• Build observability dashboards and incident remediation