Staff Engineer, ML​/AI Platform

Staff Software Engineer

Attentive® is the AI marketing platform for 1:1 personalization redefining the way brands and people connect. We're the only marketing platform that combines powerful technology with human expertise to build authentic customer relationships. By unifying SMS, RCS, email, and push notifications, our AI-powered personalization engine delivers bespoke experiences that drive performance, revenue, and loyalty through real-time behavioral insights. Recognized as the #1 provider in SMS Marketing by G2, Attentive partners with more than 8,000 customers across 70+ industries.

Leading global brands like Crate and Barrel, Urban Outfitters, and Carter's work with us to enable billions of interactions that power tens of billions in revenue for our customers. With a distributed global workforce and employee hubs in New York City, San Francisco, London, and Sydney, Attentive's team has been consistently recognized for its performance and culture. We're proud to be included in Deloitte's Fast 500 (four years running!),

Linked In's Top Startups, Forbes' Cloud 100 (five years running!), Inc.'s Best Workplaces, and the Human Rights Campaign Foundation's Corporate Equality Index!

We're seeking an accomplished Staff Software Engineer to join Attentive's Machine Learning Platform team as a high-impact individual contributor focused on building the AI and ML infrastructure that powers our AI product suite. You'll architect and build the foundational platform components that enable AI / ML engineers and data scientists to train, deploy, and serve models and agentic infrastructure with velocity, performance, and reliability  a Staff-level IC, you'll operate as a technical force multiplier, setting the technical direction for AI and ML infrastructure across Attentive's AI organization.

You'll lead through influence and technical excellence, advocating for long-term architectural progress while balancing immediate platform needs. Your work will span strategic initiatives measured in quarters and years, focusing on high-leverage decisions that enable entire teams to ship AI and ML capabilities faster and more reliably.

• Setting Technical Direction
  - Architect ML platform strategy spanning data pipelines, training infrastructure, and serving layers using cutting-edge tooling like Ray, MLFlow, Metaflow, Argo, and Spark.
• Uplevel and Innovate Core AI & ML Stack
  - Build and operate production-grade, low-latency ML serving layers with robust model lifecycle systems including champion/challenger testing, automated rollouts, versioning, and rollback capabilities.
• Define and drive Attentive's agentic stack.
• Provide ML infrastructure perspective in high-level discussions about Attentive's AI strategy spanning multiple quarters and teams.
• Mentor platform and ML engineers, actively championing team members.
• Build universal interfaces, architectures, and patterns—like data access layers and prediction serving APIs—that bridge platform capabilities with product needs to streamline high-priority ML work across the organization.

• You have the experience to know what works, what doesn't, and why in AI and ML systems.
• 5+ years focused specifically on ML Platform/MLOps, with deep understanding of gold-standard practices and best-in-class tooling.
• Proven track record of owning and building core components of ML platforms using tools like Spark, Ray, MLFlow, Kubeflow, or Metaflow.
• You've built and operated a high-throughput agentic stack (MCP / data infrastructure, context store, orchestration, and prompt layer).
• Strong expertise in Python for both batch processing and online service frameworks.
• Experience designing and operating online and offline inference systems, understanding the critical differences and tradeoffs between them.

• Design and implement inference pipelines with champion/challenger shadow testing and automated model promotion.
• Lead and scale Attentive's agentic stack from the ground up.
• Scale real-time feature streaming to handle low-latency, high-volume reinforcement learning workloads.
• Build a universal data access layer and prediction serving interface…