Building profitable PS organizations that drive customer success
I lead teams that transform enterprise AI ambitions into production-ready platforms while building profitable Professional Services organizations. With 13+ years spanning AWS, cloud architecture, and ML infrastructure, I combine technical depth with business acumen—driving utilization, reducing churn, and delivering high-value solutions for Fortune 500 clients.
Enterprise-scale implementations across AI/ML platforms, cloud infrastructure, and digital transformation.
Leading the delivery of enterprise AI platforms across EMEA for Fortune 500 companies in financial services, life sciences, and insurance. Architecting scalable MLOps infrastructure that enables data science teams to accelerate model development from research to production.
Designing and implementing agentic AI systems leveraging Retrieval Augmented Generation (RAG) for enterprise knowledge management. Building MCP servers for tool integration and developing multi-agent workflows for automated decision-making processes.
Led cloud adoption and transformation strategies for Tier-1 global banks including JPMorgan, Deutsche Bank, and UBS. Designed compliant infrastructure meeting PCI-DSS, SOC2, and GDPR requirements while enabling innovation at scale.
Architected and implemented a fully automated Infrastructure-as-Code solution for a digital insurance platform. Designed microservices architecture on Amazon EKS with multi-AZ deployment, automated CI/CD pipelines, and comprehensive observability.
Designed and implemented a high-throughput data processing architecture for real-time financial analytics. Built event-driven pipelines handling millions of transactions with sub-second latency requirements.
Transformed EMEA Professional Services into a profitable, high-impact organization delivering enterprise AI/ML platforms.
Deep dives into agentic AI, cloud architecture, and enterprise transformation.
A production deal-research team — one coordinator, four specialist sub-agents, a sandboxed DCF, a memory of prior deals, credentialed tools over MCP, and a human gate on every consequential action — built end to end on AWS with nothing leaving your account. Ships as a hands-on workshop with a starter/solution split.
A 400-line monolithic agent is easy to write and hard to trust. How I broke one into skills, sub-agents, and a sandbox — and measured that it got better. Covers agent decomposition patterns, versioned methodology skills, sandboxed compute with AgentCore, and an eval harness that grades side effects.
Claude Managed Agents and AWS Strands + AgentCore solve the same problem from opposite ends. A deep look at both architectures across seven layers — sandboxed code, multi-agent orchestration, memory, identity, observability — and a decision framework for enterprise buyers.
A personal-investing agent built by customizing AWS's Fullstack AgentCore Solution Template (FAST): three deterministic financial-tool Lambdas behind an MCP Gateway, Cedar authorization keyed off a claim a Cognito Pre-Token Lambda injects, and a Cognito-authenticated React UI. Looks through your ETFs to find true concentration, evaluates tax-aware rebalancing under German §23 EStG, and flags drawdown-triggered tranche buys — with a human gate on every trade.
A coordinator and three specialists for a fictional DTC brand, provisioned end to end with the new @aws/agentcore-cli instead of hand-rolled boto3 scripts. Introduces AgentCore Memory branching so parallel specialists can work a ticket concurrently without sharing scratch context, alongside sandboxed refund math, a live browser tool, and a Gateway that turns a Lambda and an OpenAPI spec into MCP tools.
A Drafter and a genuinely separate Critic instance turn a business's website — or a two-sentence description, when the website fights back — into a ready-to-embed intake-form config: fields, conditional logic, a sandboxed pricing formula, a quote-line mapping. No workshop mode this time — a single production-shaped build, and the write-up is as much about the wrong guesses corrected along the way (a fabricated npm package name, a Browser tool input schema that wasn't what it looked like, two missing transitive dependencies) as the architecture itself.
I'm always interested in discussing leadership roles in Solutions Architecture, AI/ML platform strategy, or enterprise transformation. Whether you're looking for a technical leader or want to explore collaboration opportunities, let's talk.