AI Full Stack Architect

AI Full Stack Architect
12 months contract
Remote (Ocassional Travel)

Tax Term: W2

About the Role
We are seeking an AI Full Stack Architect to join our AI Center of Excellence. This is a senior individual-contributor and technical leadership role for someone who has spent years in the trenches — building, shipping, and scaling AI-powered systems — and is now ready to set the architectural direction for the next generation of intelligent applications.
You will own the end-to-end design and delivery of agentic AI systems, LLM-powered platforms, and full stack applications that operate at enterprise scale. You bring deep hands-on expertise across AWS Bedrock, Python, Node.js, and modern front-end frameworks, and you know how to translate that into production-grade architecture that others can build on.
Level & Scope

DIMENSION
EXPECTATION
Seniority
Architect / Principal — senior IC with org-wide technical influence
Experience Bar
10+ years in software engineering; 4+ years in AI/ML engineering
Depth
Expert-level in at least two of: agent development, LLM integration, cloud-native backend
Breadth
Fluent across the full stack — infra, backend, AI layer, and front-end
Leadership
Drives architecture decisions, mentors engineers, sets standards
Ambiguity
Comfortable defining the problem, not just solving it
Impact
Platform-level — your decisions affect multiple teams and products

Key Responsibilities
Agent Architecture & Development
Architect and build autonomous, multi-step, and tool-using AI agents using AWS Bedrock Agents and leading agentic frameworks.
Design multi-agent orchestration topologies — including supervisor/worker patterns, parallel execution, and handoff protocols.
Establish agent design patterns: memory management, context windows, tool-call sequencing, retry logic, and failure recovery.
Define guardrails, safety layers, and responsible AI standards for all agent-based systems.
LLM / SLM Integration
Own the selection, integration, and lifecycle management of Large Language Models (Claude, GPT-4, Llama, Mistral) and Small Language Models (Phi-3, Gemma, Mistral 7B).
Design and implement RAG pipelines, prompt engineering standards, few-shot frameworks, and fine-tuning workflows.
Establish model benchmarking and evaluation criteria — balancing capability, latency, cost, and safety.
Cloud & Infrastructure (AWS)
Lead the design of AI infrastructure on AWS Bedrock — including foundation model access, Knowledge Bases, and Bedrock Agents.
Architect cloud-native backend services using Lambda, API Gateway, ECS/EKS, S3, and IAM.
Define infrastructure-as-code standards and CI/CD pipelines for AI workloads.
Full Stack Development
Build and architect full stack applications — React/Next.js front ends, Node.js backend services, and Python agent/ML pipelines.
Design streaming agent UIs, chat interfaces, and real-time AI-powered user experiences.
Own API design and integration patterns between front-end, backend, and AI layers.
Monitoring, Observability & Evaluation
Implement agent monitoring and evaluation pipelines using Fiddler AI and comparable platforms (LangSmith, Arize, Weights & Biases, Helicone).
Define and track agent performance metrics: accuracy, latency, hallucination rate, tool-call success rate, and cost-per-interaction.
Build feedback loops that drive continuous model and agent improvement.
Technical Leadership
Lead architecture reviews, design documents, and RFC processes for AI initiatives.
Mentor and upskill engineers on agent development, LLM integration, and AI best practices.
Partner with product, data, and platform teams to translate business requirements into scalable AI solutions.
Continuously evaluate emerging models, frameworks, and tooling — bringing the best to the team.

Required Qualifications
AWS & Cloud
AWS Bedrock — deep, hands-on experience building and deploying agents and models (Bedrock Agents, Knowledge Bases, foundation model APIs).
Strong working knowledge of the broader AWS ecosystem: Lambda, S3, IAM, API Gateway, ECS/EKS, CloudWatch.
Experience designing cloud-native, serverless, and containerized AI workloads