Principal Solutions Architect – AI/ML

About the role

The Strategic Accounts SSA team partners with AWS's most advanced customers to design and deploy multi-agent AI systems on Amazon Bedrock, SageMaker, and Amazon Q. As a Principal Solutions Architect specializing in agentic AI, you'll architect autonomous agent ecosystems — supervisor and broker patterns, multi-agent collaboration frameworks, and tool-using agents that plan, reason, and act across enterprise workflows with minimal human oversight.

This is a technical leadership role: you'll work directly with product teams to shape the agentic AI roadmap, build reference architectures and POCs that customers deploy in production, and set the technical bar for how AWS's broader SA community approaches agentic systems.

What you'll do

Architect multi-agent systems for enterprise customers — memory, retrieval-augmented generation, planning/reasoning loops, and tool orchestration — using Bedrock, SageMaker, and Amazon Q
Design and guide custom model training efforts, including fine-tuning foundation and open-source models (e.g., BERT, Llama family) for domain-specific agent tasks
Define and implement the model SDLC for customer engagements — data preparation, training/fine-tuning, evaluation, versioning, deployment, and monitoring/retraining loops
Establish and roll out model best-practice frameworks (evaluation rubrics, guardrails, responsible AI checks, cost/performance benchmarks) that customers and internal teams adopt as standard
Build and present proof-of-concepts and reference architectures that demonstrate measurable business impact from autonomous agent workflows
Evaluate agent reasoning performance (accuracy, cost, latency, failure modes) and translate findings into design recommendations
Partner with AWS product and service teams, feeding customer requirements into the Bedrock/agentic AI roadmap
Advise customers and internal stakeholders — from engineers to C-level executives — on agentic AI strategy, translating technical trade-offs into business terms
Create field enablement content (playbooks, workshops, reference implementations) so other Solutions Architects can replicate successful patterns
Mentor SAs and engineers on agentic system design, orchestration frameworks (e.g., CrewAI and comparable frameworks), and evaluation methodology
Stay current on agentic AI research and AWS service capabilities to keep recommendations at the leading edge

What we're looking for

15+ years of experience in technical domain such as software development, cloud infrastructure, distributed systems, or data/ML engineering
10+ years of experience in AWS with extensive usage of services like Sagemaker, S3, EKS, ECS, SQS, etc
Bachelor's degree in computer science, engineering, mathematics, or equivalent practical experience
Hands-on experience designing or implementing LLM-based or agentic AI systems (orchestration, RAG, tool use, multi-agent coordination) not just familiarity with the concepts
Hands-on experience fine-tuning and customizing foundation or open-source models (e.g., BERT, Llama, or comparable architectures) for specific use cases
Experience running the full model development lifecycle — from data prep and training through evaluation, deployment, and monitoring — and establishing best-practice standards around it
Track record of building and evangelizing technology roadmaps that guide large-scale cloud or AI transformations
Strong communicators across audiences — comfortable running workshops and technical deep-dives as well as executive/C-level conversations

Nice to have

Experience with distributed systems design at scale
Experience with large-scale workflow automation or orchestration systems
Database design and implementation experience
Demonstrated comfort whiteboarding and presenting to mixed technical/executive audiences
Experience leading customers through cloud migration, modernization, or architecture transformation