Michael Cox | Sr. AI Agentic Engineer
Zeriq IDE · Voomra · Apple · Walmart

Zeriq IDE • Voomra • Apple • Walmart

Building AI-native systems, autonomous agents, and scalable infrastructure.

Senior AI Agentic Engineer specializing in production AI systems, autonomous LangGraph workflows, RAG pipelines, and scalable marketplace platforms.

Autonomous Agents

Designing goal-driven, multi-step AI agents with LangGraph, CrewAI, and AutoGen — handling real-world production workflows.

Eval & Observability

LangSmith, Arize Phoenix, LLM-as-judge, golden sets, and adversarial red-teaming to keep agents reliable in production.

Agent Architecture

RAG pipelines, HITL gates, MCP & A2A protocols, sequential/parallel workflows, and self-correction loops built to scale.

AI-Native Infrastructure

FastAPI services, Redis kill signals, Docker/Kubernetes, Railway deployments, and GitHub Actions CI with automated eval gates.

Featured Work

Zeriq IDE

AI-native mobile IDE — LangGraph pipeline, 5 agent personas, pgvector memory

View Project

Voomra + Groundcrew

Peer-to-peer car-sharing + autonomous AI operations layer (LangGraph, FastAPI)

View Project

Technologies

Deep, production-proven expertise across the full agentic AI stack — from autonomous agent design and multi-agent orchestration to eval pipelines, observability, and AI-native cloud infrastructure. 20+ years of engineering experience spanning enterprise software, distributed systems, and now the frontier of autonomous AI systems.

Agentic Frameworks

LangGraph (State Machines & Workflows)

LangChain (Chain-of-Thought & Tools)

CrewAI (Multi-Agent Teams)

AutoGen (Microsoft)

Semantic Kernel (Microsoft)

Google ADK (Agent Development Kit)

Claude Agent SDK (Anthropic)

NeMo Agent Toolkit (NVIDIA)

LiteLLM (Multi-Model Routing)

LangFlow (Visual Agent Builder)

Core Languages

Python — Deep Proficiency

TypeScript (7yrs)

Node.js (8yrs)

JavaScript ES6+ (10+yrs)

SQL (10+yrs)

BASH (10+yrs)

HTML5 / CSS3 (10+yrs)

AI Protocols & Patterns

MCP (Model Context Protocol)

A2A Protocol (Agent-to-Agent)

ACP (Agent Communication Protocol)

RAG Pipelines

HITL (Human-in-the-Loop)

Multi-Agent Orchestration

ReAct & Chain-of-Thought (CoT)

Structured Outputs

Prompt Engineering & Injection Defense

AI Eval & Observability

LangSmith

LangFuse & Braintrust

Arize Phoenix (Self-Hosted)

OpenTelemetry (OTel) Auto-Instrumentation

Golden Sets & Offline Regression Suites

LLM-as-Judge Patterns

Online A/B & Shadow Evaluations

Adversarial Red-Teaming

PII Leakage & Prompt Injection Testing

Vector & Data Storage

pgvector via PostgreSQL

OpenSearch

Redis (Kill Signals, Rate Limiting)

Kafka (Event Streaming)

Redshift & BigQuery

MongoDB (6yrs)

Apache Airflow (ETL Pipelines)

SageMaker / Data Lakes

Infrastructure & Cloud

FastAPI & gRPC Services

Docker w/ Kubernetes

AWS · Azure · GCP · Railway

GitHub Actions CI/CD (Eval Gates)

Terraform & ArgoCD

Pydantic & Pydantic Settings

React.js / Next.js (Frontend)

AI-Assisted Dev (Claude Code, Cursor, v0)

Frontier Models: OpenAI, Anthropic, Gemini, Bedrock

Ready to build autonomous AI systems that work? Let's talk.

Portfolio

I'm proud of what I have done. I'm ready for the next big challenge. Yours!
Take a look at my most recent projects.

Works match criteria

Project Details

    SafeRyde, Inc

    Lead Software Engineer
    Date : November 2015
    Categories : Applications

    The Challenge

    With the up raise in peer to peer transporation services like Uber and Lyft many traditional transporation companies are slowing going out of business. With peer to peer comes risk, accountability, and safety concerns which many customers are not willing to trust. SafeRyde discovered a need and a niche in the peer to peer transporation industry for kids. The big players in the industry have neglected to fill this niche both with insurance coverage and safety for peer to peer transporation for kids.

    The Solution

    Build infrastructure with scalability in mind. Amazon AWS Cloud utilizing the complete LAMP Stack using both MYSQL Database and MondoDB. Backend language is PHP and Frontend technologies include JavaScript, Jquery, AngularJS, HTML5, and CSS3. Mobile application build with Phonegap framework for ease portability to both Android and iOS platforms. Both backend and frontend codebase is MVC. With enhanced API we can easily integrate with third party systems for partnerships and integrations.

    What makes are technology and service unque is our safety features, security, and a nichy market. With our realtime in vehicle video streaming, live realtime tracking, realtime driver analytics, enhanced background, drug testing, and status alerts gives our customers the feeling of security and being in two places at the same time.

    The Result

    Scalable & hightly tested Web Application, API, and Native Hi-bread Mobile Apps.

    Anytimewallet.com

    Lead Software Engineer
    Date : July 2014 - October 2015
    Categories : Marketing Site & LMS

    The Challenge

    This project required alot of planning because of the nature and risk of attacks on financial sites. By collecting consumer sensitive data and financial information we had to implement some of the highest security features in the industry. In the requirements, planning, and design phase of this project we needed the development to start with security in mind.

    The Solution

    We implemented oneway password encryption, disabled all known PHP exploit volunerbilities as will as firewall and database access restriction beyond the usual standards. Developing on a tested proven strong full stack PHP Framework allowed us to utilize tested libraries to help in the development process and not try to re-create the wheel on some feature like authentication and data sanitization.

    The Result

    With the collaboration from System Admins, DBA's, and developers we was able to achieve all goals and requiremnts withing budget and deadline. Resulting in a product that is security reliable, and secure. We implemented security features on the frontend, backend, database, and at the infrastructure level. LMS system is running smooth in production.

    Ticketmaster/Livenation

    Security Console, API Integrations
    Date : November 2013
    Client : Livenation
    Categories : Administration Console, Security, API Integrations

    The Challenge

    Ticketmaster was acquired by Livenation and that fueled demand for data sharing and integrations. Challenge was to create a robust, fast, secure optimized API Service for Ticketmaster.com, BigChampagne, Livenation.com, and rdio app.

    The Solution

    Phalcon PHP MVC framework was used to develop a high performance API Web Service that will handle a high demand API responses. Phalcon was used because it C binary and runs as a php module. I also created a Codeigniter authentication API Server, as well as a API key management application.

    Cox Wifi Project

    Wifi Internet Captive Portal
    Date : Dec 2012
    Client : Cox Communication Inc..
    Categories : Application Development, Security, API Development, API Integration

    The Challenge

    Cox Communication is a cable service company. Cox enter an alliance with other cable companies to provide wifi service to visiting customers from other cable services. A internet captive portal was needed to capture user credentials and integration with Broadhop a Cisco company.

    I was part of a time of 8 developers to role out this application using PHP Symphony MV Framework. Integrating with Broadhop via Soap API Web Service.


    I'm ready for the next challenge. Yours!

    Contact Me

    • Michael was a warrior trying to get our item and configuration overview releases out on time. Great Job!

      Michael Cox / Team Review Walmart

    • Michael has done a really great job with all his stories, helping out with bug fixes, and is always keeping a positive attitude. Crushed It!

      V. Matusevich / Manager Apple

    • Michael Cox delivered our web application successfully within budget and deadline. His overall experience and skills helped us choose the correct infrastucture and technologies to use that would allow us to expand in the near future.

      Tim Mcgraw / President Voomra

    About Me

    I'm a Sr. AI Agentic Engineer with 20+ years of software engineering experience, now focused entirely on autonomous AI systems. I started coding as a teenager in Silicon Valley during the dot-com era and have spent my career at the intersection of distributed systems, enterprise architecture, and cutting-edge product development — working with organizations like Apple, Walmart, and Ticketmaster/Livenation before turning my focus to the frontier of agentic AI.

    Today I specialize in designing and shipping production-grade multi-agent systems: autonomous LangGraph workflows, multi-agent orchestration with CrewAI and AutoGen, RAG pipelines, Human-in-the-loop (HITL) gates, and LLM guardrail layers. My most recent work is Groundcrew — a fully autonomous AI operations platform built for Voomra that runs four specialized LangGraph agents behind a FastAPI service, backed by Redis kill signals, Pydantic-typed I/O, LangSmith observability, and a CI eval gate (57 test cases, ≥85% pass rate required to deploy).

    I approach agentic systems with a deep respect for reliability, observability, and security. That means adversarial red-teaming for prompt injection and PII leakage, golden-set eval suites before any model upgrade, and OTel auto-instrumentation so every agent run is fully traceable from root span to tool call to LLM response.

    Other highlights of my background:

    • Deep proficiency in Python alongside TypeScript/Node.js; fluent in MCP, A2A, and ACP inter-agent protocols.
    • Built and shipped autonomous workflows using LangGraph, CrewAI, AutoGen, Semantic Kernel, Google ADK, and Claude Agent SDK.
    • Experienced with AI eval stacks (LangSmith, LangFuse, Braintrust, Arize Phoenix), LLM-as-judge patterns, and online A/B & shadow evaluations.
    • Cloud infrastructure across AWS, Azure, GCP, and Railway; Terraform, ArgoCD, Docker, Kubernetes, and GitHub Actions CI/CD with automated eval gates.
    • Strong record of building and leading engineering teams at enterprise scale — Apple, Walmart, Ticketmaster/Livenation, and early-stage startups alike.

    Outside of engineering, I ride Harley cruisers, play basketball, and spend time with my family. Ambitious but practical — I don't over-promise, and I always deliver.

    Contact

    I would love to learn about your project.
    Write to me or call me

    Drop me a line

    United States

    Michael Cox

    Contact Hours

    • Mon - Fri : 9am - 5pm