Staff AI Builder

robotsandpencils; United States; Remote; USD 180,000 - 250,000; Full time

Job summary

robotsandpencils is seeking a senior AI systems architect to lead the design and delivery of production GenAI and machine-learning platforms. You will own model workflows from concept through integration, deployment, and ongoing reliability, guiding technical decisions on a cross-functional team and shaping engineering practices for secure, scalable AI products.

You will design agent workflows, build retrieval-augmented generation pipelines, develop prompts and observability, and maintain

Qualifications

• Six years of professional software engineering experience, including production GenAI or LLM systems.
• Experience with agentic workflows, RAG, prompt design, and evaluating and monitoring LLM applications.
• Strong Python and Node.js skills, including building REST APIs.
• Hands-on AWS AI and application services (Bedrock, Lambda, DynamoDB, S3, SQS, EventBridge, Step Functions).
• Knowledge of production reliability patterns and cloud-native deployment.
• Ability to own ambiguous integration work and explain technical tradeoffs.

Responsibilities

• Design agent workflows, including tool calling and single- or multi-agent orchestration.
• Build retrieval-augmented generation pipelines with chunking, embeddings, vector search, reranking, and refresh handling.
• Develop prompts, integrations, evaluation methods, and observability for production AI systems.
• Build and maintain services and APIs using Python, Node.js, and cloud infrastructure; support event-driven and serverless workflows.
• Design data models for conversation state and agent memory, and add resilience through retries, circuit breakers, and fallback models.
• Lead architecture discussions, mentor engineers, and own features through release, debugging, and production hardening.

Skills

GenAI / LLM experience

Agentic workflows design

REST API development

Python

Node.js

AWS services

Tools

AWS Bedrock

Lambda

DynamoDB

S3

SQS

EventBridge

Step Functions

Job description

Role overview

Lead the architecture and delivery of production AI and machine-learning systems, taking ownership from model and workflow design through integration, deployment, and ongoing reliability. You’ll guide technical decisions on a cross-functional team, solve complex integration challenges, and help establish engineering practices for secure, scalable AI products.

Responsibilities

• Design agent workflows, including tool calling and single- or multi-agent orchestration, choosing patterns to fit the problem.


• Build retrieval-augmented generation pipelines with chunking, embeddings, vector search, reranking, and refresh handling.


• Develop prompts, integrations, evaluation methods, and observability for production AI systems.


• Build and maintain services and APIs using Python, Node.js, and cloud infrastructure; support event-driven and serverless workflows.


• Design data models for conversation state and agent memory, and add resilience through retries, circuit breakers, and fallback models.


• Lead architecture discussions, mentor engineers, and own features through release, debugging, and production hardening.

Requirements

• At least six years of professional software engineering experience, including production GenAI or LLM systems.


• Practical experience with agentic workflows, RAG, prompt design, and evaluating and monitoring LLM applications.


• Strong Python and Node.js skills, including experience building full-stack applications and REST APIs.


• Hands-on experience with AWS AI and application services, such as Bedrock, Lambda, DynamoDB, S3, SQS, EventBridge, or Step Functions.


• Knowledge of production reliability patterns and containerized, cloud-native deployment.


• Ability to take ownership of ambiguous integration work and explain technical tradeoffs in depth.

Nice to have

• Experience with Bedrock Agent Core, MCP tool integration, or state-machine workflow orchestration.


• Background in regulated or high-stakes domains, or experience running LLM observability tools in production.