Find the adoption loop
Start with the workflow, decision points, data constraints, owners, and the behavior change needed for AI to move the outcome.
John Carroll
Forward-Deployed AI · Agentic Systems · Team Builder
I build AI systems and teams from prototype to production.
For organizations turning high-value workflows into shipped AI products: forward-deployed engineering, enterprise RAG, agentic automation, multimodal field tools, and cloud-native platforms.
How I Operate
AI from prototype to adoption.
I help organizations turn high-value workflows into AI systems that people can trust, adopt, and keep using in production.
Over the past decade, I've designed and delivered AI and software systems across healthcare, retail, manufacturing, aerospace, and public safety. My work focuses on agentic AI, RAG, enterprise search, AI platforms, cloud architecture, and modern software engineering, with an emphasis on secure, governed, reliable systems at scale.
The main reason AI work fails is adoption. I push the needle by building close to real workflows, aligning stakeholders, staffing the right delivery model, and making sure the system fits how people actually operate.
My background started in software and electrical engineering, from embedded systems and industrial technology to cloud platforms and enterprise software. That foundation shapes how I approach AI today: as an engineering discipline grounded in reliability, scalability, and measurable business outcomes.
I like operating at the intersection of strategy and execution: shaping technical direction, building platforms, mentoring engineers, advising leaders, and helping teams turn emerging AI capabilities into useful products.
Build the useful system
Prototype quickly, wire real interfaces, test with messy inputs, and keep the experience close to the people using it.
Make it production-minded
Harden retrieval, evaluation, guardrails, observability, cost, latency, and deployment until the system can be trusted.
x
x
$200M+
AI-related revenue influenced
Microsoft AI Global Black Belt
Helped enterprise customers move AI opportunities from experimentation into funded production work.
$25M
annual efficiency savings
Boeing
Built analytics and value-stream systems tied to recurring operational savings across major aerospace programs.
50+
AI talent hired / staffed
Forward-deployed team building
Built and staffed AI delivery capacity across engineers, data scientists, TPMs, and ontologists, including forward-deployed pods and playbooks.
Adoption & Operating Leadership
Adoption is the operating model.
The work is not only building the prototype. It is shaping the team, stakeholder loop, workflow fit, enablement path, and production model that make people actually use it.
Build the team
Hiring, staffing, onboarding, role definition, and engineering standards for forward-deployed AI work.
Shape the adoption model
FDE playbooks, delivery structure, stakeholder rhythms, intake patterns, enablement paths, and repeatable production delivery.
Advise the business
Leadership advisory, AI platform direction, build-vs-buy decisions, governance needs, and executive-facing demos.
Ship production AI
Evaluation, monitoring, security, reliability, cost, latency, deployment, and operational readiness for systems people can trust.
AI Lane
Agentic AI, RAG, enterprise search, multimodal assistants, and AI platforms for production workflows at scale.
Domains
Healthcare, retail, manufacturing, aerospace, public safety, industrial systems, and enterprise operations.
Delivery Range
Strategy, staffing, architecture, prototypes, production systems, cloud platforms, full-stack apps, and engineering mentorship.
Operating Model
FDE pods, playbooks, stakeholder loops, evaluation, governance, and production delivery that teams can repeat.
Enterprise AI Customer Footprint
Organizations I've supported through enterprise AI.
A representative set of organizations I supported, advised, partnered with, or engaged through Microsoft AI GBB and enterprise AI work, plus state and local public-sector teams.
Retail / Consumer
CPG / Food / Distribution
Aerospace / Manufacturing / Tech
Finance / Services / Sports / Other
Experience
Work history, education, and recognition.
The career arc behind the work: enterprise AI, forward-deployed teams, aerospace analytics, public safety SaaS, industrial systems, consulting, and the teaching foundations that shaped how I build.
Work History
Roles, systems, and outcomes.
A timeline of the teams, products, operating models, and production systems behind the portfolio.
01 / Nov 2025 - Present
Cigna / Evernorth
Senior Principal Architect
xLead applied AI architecture and forward-deployed engineering in the AI Enablement Office, helping establish the FDE model, delivery playbooks, hiring guidance, and production AI operating structure.
02 / Jan 2011 - Present
Freelance
Full-Stack Developer and Consultant
Partner with clients to design, build, and host scalable web applications; recently served as sole developer for a stealth startup, shipping two MVPs to production in May 2025 after a two-month build and helping generate six-figure profits and new contracts.
03 / Sep 2023 - Nov 2025
Microsoft
Senior Solution Engineer
Trusted advisor and architect for Fortune 500 AI adoption, influencing $200M+ in AI-related revenue through technical strategy, production-grade agentic systems, reusable accelerators, and scalable deployment patterns.
04 / Jun 2019 - Sep 2023
Boeing
Senior AI/ML Software Engineer
Delivered AI and analytics systems for high-profile programs, including work credited with $25M in annual efficiency savings.
05 / Jun 2016 - Jul 2019
Hexagon
Full-Stack Software Engineer
xBuilt cloud-native public safety systems spanning IoT, command-and-control, video analytics, and DevSecOps delivery.
06 / Aug 2015 - May 2016
Auburn University
Computer Science Teaching Assistant
Supported computer science instruction and student learning while completing the software engineering foundation behind later production work.
07 / Jan 2013 - May 2015
Neptune Technology Group
Software Engineer, Co-op Program
Completed three software and electrical engineering co-op rotations in industrial technology before moving into full-stack and AI systems work.
Recognition
Work that got noticed.
Jan 2026
Fastest Production Release
Lead Engineer / Cigna
Recognized for the fastest production release in the AI Enablement Office.
Aug 2025
Microsoft Global Hackathon
Judge / AI Agents Category
Served as a judge for agentic AI submissions across the global hackathon.
Apr 2024
AI Global Black Belt Hackathon
Lead Developer / 1st Place
Led development for the winning Microsoft AI Global Black Belt entry.
Feb 2023
Alabama 20th Annual STARS Award
Team Lead / Boeing
Predictability and Stability recognition tied to $25M in annual efficiency savings.
Education
Formal foundations, still compounding.
Two completed degrees in software engineering and executive business, plus a graduate AI specialization about to begin.
Georgia Institute of Technology
Master of Science in Computer Science, AI Specialization
Expected Start: August 2026 / Incoming
Incoming graduate computer science study at a top-30 global CS program and No. 9 U.S. public university, connected to applied AI systems and production engineering.
Quantic School of Business and Technology
Executive Master of Business Administration
March 2025 - July 2026 / Completed
Completed the Executive MBA with HelixGuard, a clinical AI governance and decision-support platform, as the 2026 capstone project. The program strengthened the strategy and operating-model context behind translating technical systems into useful outcomes.
Auburn University
Bachelor of Science in Software Engineering
August 2011 - May 2016 / Completed
ABET-accredited software engineering foundation, including a one-year co-op program, hands-on CS teaching experience, and coursework in complex software systems.
Featured Systems
Demo-rich AI work, framed by outcomes.
A curated set of systems spanning clinical AI governance, agentic reasoning, RAG, realtime voice, video understanding, and document automation.
Govern → Embed → Measure
HelixGuard (formerly ClinicalOS)
Clinical AI governance and decision support across purpose-built clinician, leadership, and compliance surfaces backed by a shared audit layer.
ReactTypeScriptMUIClinical AIGovernance
View system
Real-time disruption reasoning
Supply Chain Reasoning Engine
Adaptive reasoning workflows for disruption response, scenario planning, and resilient supply chain decisions.
TypeScriptPythonAISupply Chain
View system
Voice, vision, and grounded retrieval
Field Technician Assistant
Hands-free multimodal RAG for technicians working from complex technical documentation in the field.
Azure AI SearchOpenAIVoiceRAG
View system
Realtime GPT-4o voice UX
Coffee Chat Voice Assistant
Realtime voice ordering system with live transcription, natural conversation, and streaming order updates.
ReactPythonRealtimeVoice
View system
Chat over video evidence
Video Analysis with LLMs
AI vision workflow that segments video, extracts evidence, and enables interactive chat over visual content.
PythonStreamlitVisionFFmpeg
View system
Unstructured docs to usable JSON
Document Extraction to JSON
Schema-guided document extraction pipeline that turns scanned and digital PDFs into structured data.
PythonOCRVisionExtraction
View systemWhat I Offer
Services
AI Workflow Automation
I find the workflow, scope the automation opportunity, and build the useful system: document processing, data extraction, tool integration, and production-minded internal apps.
AI Operating Model Advisory
I help teams decide how AI work should actually run: intake, staffing, governance, platform choices, delivery rhythms, evaluation, and production ownership.
Forward-Deployed AI Team Design
I help organizations design and staff forward-deployed AI pods that can move from ambiguous business problems to working systems without getting trapped in demo land.
AI Coaching & Enablement
Practical enablement for leaders and teams adopting AI tools, agentic workflows, RAG systems, and production-minded engineering patterns.
What I Build
Practical AI systems for messy workflows.
Strategy is useful when it turns into working software, repeatable teams, and operating models that can survive production.
Agentic Workflows
Planning, tool use, orchestration, memory, guardrails, and human-in-the-loop review for workflows that need more than a chatbot.
Enterprise RAG and Search
Ingestion, chunking, indexing, semantic and hybrid search, grounding, citations, evaluation, and reliability monitoring.
Voice and Multimodal Assistants
Realtime speech, image analysis, video understanding, and grounded assistant experiences for hands-free or visual workflows.
AI Team & Operating Model Design
FDE pod design, hiring guidance, delivery playbooks, stakeholder rhythms, governance, and engineering standards for production AI.
Evaluation and Guardrails
Benchmarks, safety controls, cost/latency tradeoffs, quality monitoring, and review loops that keep AI systems accountable.
Cloud-Native Full-Stack Systems
React, Next.js, Python, APIs, queues, data pipelines, observability, CI/CD, and cloud architecture stitched into real products.
Technical Systems Map
How the pieces fit together.
The stack changes by problem, but the operating model is consistent: useful data, grounded retrieval, agentic action, evaluation, and production delivery.
Data Sources
- Enterprise systems
- Documents & files
- APIs & services
- Streaming data
Ingest & Process
- ETL pipelines
- Chunking & blending
- Embeddings
- Metadata
Knowledge Layer
- Vector stores
- Knowledge graphs
- Hybrid search
- Grounding
Agentic Layer
- Planning
- Tool use
- Memory
- Evaluation
Delivery
- Web apps
- APIs
- Voice UX
- Integrations
Guardrails, governance, observability, and evaluation run across every layer.
Let's Connect
Have an AI workflow, an interesting idea, or just want to connect?
Reach out for services, collaboration, AI adoption questions, prototypes, architecture, or just because something here sparked a thought. I am always up for a good build conversation.
Atlanta, GA