Principal Software Engineer, AI

Salesloft is building an AI-powered Predictive Revenue System for enterprise revenue teams. The Principal Software Engineer, AI will set the technical direction for AI systems, architect production-scale systems involving LLMs, machine learning models, and agentic workflows, and establish standards for reliable, measurable, and trustworthy AI. The role also leads complex AI engineering initiatives and mentors senior engineers across the organization.

Responsibilities:

• Set the technical architecture and direction for AI systems across the platform — model serving, LLM-powered features, agentic workflows, and the infrastructure beneath them
• Design production-grade systems that are reliable and observable despite being built on non-deterministic, probabilistic AI components
• Architect evaluation frameworks, guardrails, and quality systems that make AI behavior safe, measurable, and trustworthy at enterprise scale
• Lead the most complex AI engineering problems end to end — from research-adjacent prototyping through hardened production delivery
• Establish engineering standards and best practices for building AI-native systems, and mentor senior engineers across the org
• Keep the platform at the frontier by evaluating and integrating advances in foundation models, agent frameworks, and the modern AI stack

Requirements:

• Extensive experience building and architecting production software systems at scale, with significant depth in AI/ML systems
• Hands-on experience building production systems on top of LLMs or machine learning models — including designing for the challenges of non-deterministic behavior
• Deep expertise in distributed systems, model serving, and the infrastructure required to run AI reliably at scale
• Proven experience designing evaluation frameworks, guardrails, or quality systems for AI-driven features
• A track record of setting technical architecture and standards across teams and mentoring senior engineers
• Strong command of a major programming language (e.g., Python, Java, or Go) and modern cloud infrastructure (AWS or GCP)
• Hands-on experience with agentic systems — designing and hardening AI agents that take autonomous action in production
• Depth in the modern AI stack — retrieval-augmented generation, vector stores, agent frameworks, model orchestration, or evaluation tooling