Staff Software Engineer, Model Infrastructure

Lead design and development of Harvey's Model Infrastructure platform powering all AI requests, including unified model controller, intelligent routing, multi-provider integrations, observability, and capacity management for high reliability, low latency, and efficiency. Requires 7+ years building large-scale distributed systems with strong programming and leadership skills; AI/LLM infrastructure experience preferred.

About the job

What You’ll Do

Lead the design and implementation of Harvey's Model Infrastructure platform.
Build systems to ensure high availability, low latency, and operational excellence for AI inference.
Design and improve Harvey's Unified Model Controller (UMC) and Model Selector platform to automatically detect model degradations and intelligently route traffic based on reliability, latency, quality, compliance, and cost.
Develop systems for model provisioning, capacity management, failover, and traffic engineering across multiple AI providers.
Integrate new model providers and maintain provider APIs and SDKs, enabling Harvey to rapidly adopt emerging frontier models.
Improve observability through health dashboards, alerting, token usage analytics, cost reporting, and end-to-end telemetry.
Partner with Product Engineering to support model launches, experimentation, and proactive monitoring of production AI workloads.
Drive infrastructure efficiency through capacity planning, utilization optimization, and cost visibility.
Collaborate with AI Research to build the infrastructure foundation for future model evaluation, training, and deployment.
Lead cross-functional technical initiatives and mentor engineers across the organization.

What You’ll Build

Model Reliability & Operations: Model health monitoring, automated failover and recovery, capacity provisioning, operational tooling and incident automation, Unified Model Controller (UMC), policy-based model routing, Intelligent Model Selector, traffic management, reliability and latency optimization.
Provider Platform: Multi-provider architecture, API and SDK integrations (OpenAI, Anthropic, Azure OpenAI, Fireworks, Baseten, and future providers), rapid adoption of new frontier models.
Observability & Cost Platform: Token usage analytics, cost attribution, latency and reliability dashboards, capacity forecasting, utilization optimization.
AI Platform Foundation: Infrastructure supporting model evaluation, model deployment and operations, future model training platform, agent infrastructure and CcaaS.

What You Have

7+ years of software engineering experience building large-scale distributed systems.
Experience designing and operating highly available production services.
Strong programming skills in Go, Java, Python, Rust, or C++.
Deep understanding of distributed systems, cloud infrastructure, networking, and observability.
Experience leading technical projects across multiple engineering teams.
Ability to balance long-term architecture with pragmatic execution.
Strong communication and collaboration skills.
Passion for building foundational platforms that enable other engineering teams.

Nice to Have

Experience with AI infrastructure, LLM serving, or machine learning platforms.
Experience with model routing, inference gateways, or policy-based serving systems.
Experience working with OpenAI, Anthropic, Azure OpenAI, Fireworks, Baseten, or open-source LLMs.
Experience with Kubernetes, cloud infrastructure, and service mesh technologies.
Experience with large-scale observability and SRE best practices.
Experience with data infrastructure technologies such as Kafka, Spark, Flink, Airflow, or Iceberg.
Familiarity with GPU infrastructure or model training platforms.