Steven Mays
AI Engineer.
I build production AI on backbone systems that don't break. Started in backend a decade and a half ago — APIs, event-driven systems, serverless on AWS — and over the last few years moved that depth into LLM agents, MCP servers, and agentic pipelines that automate real work.
01 / Highlights
- 01 Production LLM agents at scale across compliance, brand protection, and marketing
- 02 Agentic pipelines & state-machine orchestration with human review gates
- 03 MCP servers & data connectors making internal data accessible to Claude
- 04 Fault-tolerant event-driven systems — sagas, circuit breakers, idempotency
- 05 Cloud-native serverless migrations cutting infrastructure cost by 50%
- 06 Marketing automation system that moved $1M+ in annual revenue
02 / Stack
- AI / LLM Engineering
- Claude (Anthropic SDK), OpenAI, MCP (Model Context Protocol) servers, agent orchestration, prompt engineering & versioning, agentic state machines, human-in-the-loop, LLM cost tracking, RAG, structured outputs, tool/function calling, Claude Skills, Claude Code
- Languages
- TypeScript, JavaScript, Go, C#, SQL, GraphQL, Python
- Cloud / Infra
- AWS (Lambda, API Gateway, SQS, SNS, Kinesis, DynamoDB, S3, RDS, CloudFormation, CloudWatch, ECS, App Runner, Amplify), Serverless Framework, AWS SAM & CDK, Docker, Kubernetes
- Databases
- PostgreSQL, MongoDB, MySQL, DynamoDB, SQL Server, Snowflake, Redis, Elasticsearch
- Architecture & Tooling
- Microservices, event-driven (NATS, RabbitMQ), REST/GraphQL, CI/CD, GitHub Actions, Semantic Release, Segment, Airtable
03 / Education
B.S. in Management / Computer Information Systems · Park University · 2011
Always up for a conversation.
AI agents, MCP servers, hard backend problems, or just trading notes — happy to dig in.