🛠 Tech Stack

💼 About This Role

You'll architect the platform that powers a swarm of hundreds of specialized AI services for travel and expense management. You'll own the runtime environment, optimize inference on SageMaker, and build deterministic shells around probabilistic LM outputs. This role offers the challenge of moving beyond traditional MLOps to manage a factory of Language Models with high-density inference optimization.

🎯 What You'll Do

• Build and own the runtime environment for 100+ specialized AI services.
• Design and implement SageMaker Multi-Model Endpoints and Inference Components.
• Build deterministic shells around probabilistic LM outputs for reliability.
• Implement LLM-as-a-judge patterns and automated benchmarking for semantic drift.

📋 Requirements

• 5+ years in SRE, Platform Engineering, or MLOps with 2+ years deploying LLMs/SLMs.
• Deep hands-on expertise with AWS SageMaker Multi-Model Endpoints and GPU-backed instances.
• Proven experience with Small Language Models (Mistral, Llama 3, Phi) and PEFT strategies like LoRA.
• Strong proficiency in Python and Terraform.

✨ Nice to Have

• Experience with Docker, Kubernetes (EKS), or AWS ECS/Fargate.
• Familiarity with Snowflake and Vector Databases.
• Experience building CI/CD pipelines for non-deterministic software.