DaveThibault
Building anything
we can envision.
Senior technical leader with deep cloud architecture, software engineering, devops, and MLOps experience. Advising C-suite executives to hands-on developers to accelerate AI-powered products to market on AWS.
About
From Architecture to AI Innovation
I'm a Senior Applied AI Architect at Amazon Web Services, where I help customers build whatever we can envision to deliver value for their businesses. My work spans the full spectrum from advising C-suite executives on AI strategy to writing production code alongside development teams.
With deep experience in cloud architecture, software engineering, DevOps, and MLOps, I specialize in accelerating AI-powered products to market. I focus particularly on Generative AI, Retrieval-Augmented Generation (RAG), and conversational AI systems built on AWS.
I'm also a technical writer and open source contributor, sharing practical knowledge through detailed articles on topics like Graph RAG, intelligent document processing, and multi-tenant AI application architectures.
Highlights
Career Milestones
Key achievements and contributions across the AI and cloud landscape.
Two World Firsts
AWS’s First Anthropic Claude Champion
Selected as the very first Anthropic Claude Champion at AWS at the end of 2023 — for learning faster, diving deeper, and enabling more builders than anyone else in AWS during the 2023 generative AI rush. Sent over 7,000 Slack messages that year, the vast majority answering AI questions for Amazonian builders across channels with thousands of members.
The Proof of Concept Behind SAP Joule
In spring 2024, was the first to demonstrate that a RAG system could pass an SAP certification test, when SAP engaged AWS and Dave had the opportunity to prove the concept that would become their Joule Copilot. Used ~900 pages of SAP training docs with Claude 2.1 to bring the score from 38% to 74% (60% passing). In 2025 SAP moved Joule’s inference to Amazon Bedrock and Claude Sonnet 4+ with exponential growth following.
Thought Leadership
The Paths to Generative AI Profits
Published a 2023 framework ranking the three paths to gen AI profits — enablement, lowering COGS, and model creation — ordered by speed to market and ROI. A year later, the predictions proved remarkably accurate: enablement companies profited first, while model companies remained in the red due to the cost and timelines of path three.
Multi-Tenant Full-Stack RAG Application
Built an open-source multi-tenant RAG stack on AWS featuring document collections before Bedrock Knowledge Bases existed, graph RAG before Bedrock KBs added graph RAG, and multi-tool/multi-context agentic orchestration since 2024.
Expertise
What I Build & Advise On
Deep technical expertise across the AI and cloud stack, from ideation through production deployment.
Generative AI & LLMs
Building production applications with large language models, including Amazon Bedrock, Claude, and multi-modal AI systems. Prompt engineering, fine-tuning, and model evaluation.
RAG & Graph RAG
Designing retrieval-augmented generation systems for enterprise use cases. Multi-tenant vector databases, knowledge graphs, entity extraction, and semantic search architectures.
Cloud Architecture
End-to-end AWS solution design using CDK, serverless, containers, and managed services. Security-first, cost-optimized, and production-ready infrastructure as code.
Intelligent Document Processing
OCR with LLMs, vision-aware RAG, document classification, table extraction, and multi-format processing pipelines for enterprise document workflows.
Full-Stack Development
React, Next.js, TypeScript frontends paired with Python, Node.js backends. API design, authentication, real-time streaming, and multi-tenant SaaS patterns.
MLOps & DevOps
CI/CD pipelines, infrastructure as code, containerization, model deployment, monitoring, and operational excellence for AI/ML workloads at scale.
Writing
Technical Articles on Medium
Practical, hands-on articles covering real-world AI architectures and implementation patterns.
Graph RAG Part 1: What Is It, When You Need It, & How to Do It
A deep dive into graph databases, knowledge graphs, and how they complement vector-based semantic search in RAG applications. Covers node/edge concepts, JSON graph formats, and when graph RAG is the right pattern.
Graph RAG Part 2: Multi-Tenancy, Semantic Search, and Multi-Context Retrieval
Integrating graph databases with multi-tenant vector database systems. Using LLMs to create graph queries from user prompts and available schema information for powerful contextual retrieval.
OCR and Intelligent Document Processing with LLMs
One of the most popular GenAI use cases in 2024. Covers vision-text compression, layout-aware reasoning, tiered fidelity approaches, and practical architecture for processing documents at scale.
Projects
Featured Work
Building real-world applications that demonstrate production-grade AI patterns.
Multi-Tenant Full-Stack RAG Application
A comprehensive, production-ready demo application published on AWS Samples. Features multi-tenant document collections, prompt management, LLM-based OCR, entity extraction, graph queries, and automatic orchestration of incoming prompts across available document collections for conversation context.
Connect
Let's Build Something Together
Whether you're looking for AI strategy guidance, architecture advice, or want to discuss a technical challenge, I'd love to connect.
Reach me on LinkedIn
The best way to get in touch is through LinkedIn. Send me a message and let's discuss how I can help with your AI and cloud initiatives.
Connect & Message on LinkedIn