W. Allan Edwards
AI/ML Architect · Hands-On Solutions Architect · Senior Software Engineer
Professional Summary
AI engineer and enterprise solutions architect with 20+ years building production systems for Fortune 500 clients. Ships applied LLM solutions end to end—integrating Anthropic Claude (via API and AWS Bedrock) into production .NET and Python systems for code conversion, document analysis, and conversational reporting, and pairing it with classic ML (PyTorch, ML.NET, scikit-learn) for deterministic, model-driven scoring. Deep hands-on experience across RAG, prompt engineering, structured-output orchestration, MCP server integration, and distributed ML training (LSTM/DeepAR on Ray, Anyscale, and EKS). Led teams of up to 160 developers across on-shore and off-shore engagements. Strong on cloud platforms (Azure, AWS, GCP), .NET/C#, Python, Go, and Kubernetes. Started career at Microsoft at age 19; since then delivered solutions for Microsoft, IBM, Koch Industries, Fiserv, American Airlines, Verizon, and SiriusXM.
Core Competencies
AI & Machine Learning
- LLMs · GenAI · Agentic AI · AI Agents
- Claude (API + AWS Bedrock) · Azure OpenAI · Vertex AI
- RAG · Vector DBs · Embeddings · Prompt Engineering
- LangChain · LlamaIndex · MCP · Evals · Fine-Tuning
- PyTorch · TensorFlow · ML.NET · scikit-learn
- LLMOps · MLOps · SageMaker · Hugging Face
Cloud & Infrastructure
- Azure · AWS · GCP
- Kubernetes · Docker · Terraform
- Serverless · Microservices
- CI/CD · DevOps · GitOps
Development & Data
- .NET/C# · Python · Go
- React · Angular · Node.js
- SQL Server · PostgreSQL · BigQuery
- Databricks · Synapse · Hadoop
Featured Portfolio Project
Grade My Investments (GMI) — AI-Powered Stock Analysis & Rating Platform
- Full-stack SaaS platform built on Azure using AI-first development practices (Claude, Copilot) with Blazor web frontend, .NET backend, and MAUI cross-platform mobile app
- Architected a hybrid AI/ML pipeline: ML.NET drives deterministic, portfolio-scale Buy/Sell/Hold grading with a consistent methodology, while Claude (Haiku, Sonnet, Opus) handles conversational analysis, natural-language report interpretation, and longitudinal comparison—keeping scoring reproducible while delivering AI-generated narrative
- Automated ingestion of financial data (balance sheets, income statements, cash flow) via licensed FMP API with Stripe server-side billing and Google/Apple OAuth authentication
- Engineered LLM reliability and cost controls for pay-as-you-go economics: model tiering across Haiku/Sonnet/Opus, retry/backoff, caching, and request staggering for rate-limit handling; scale-tested with 200–500 virtual users
Tech:
C#, .NET 8, Blazor, MAUI, Azure App Service, Claude API, Stripe, ML.NET, Google OAuth, PKCE
Professional Experience
Senior Consultant / Solutions Architect
Independent consultancy delivering enterprise architecture, cloud modernization, and ML/AI solutions.
Healthcare Interoperability Platform — AI-Driven Cloud Migration Engine (AWS→GCP, FHIR) — Senior AI Engineer
- Co-architected a TypeScript console application driving Anthropic Claude via AWS Bedrock to run a standardized, repeatable migration analysis across 350+ code bases—producing consistent, deterministic AWS→GCP conversion output at a scale no manual review could match
- Engineered a multi-pass agentic workflow: each pass generated documentation re-fed into Claude to port the FHIR service-client, service-API, and core source bases—targeting Claude specifically for code-reasoning and summarization quality
- Built custom MCP servers for the conversion pipeline and used Claude to design a Pub/Sub abstraction layer (Python + TypeScript, fully unit-tested) decoupling Pub/Sub from Kafka, plus OAuth2 (Ping One→GCP) and eventing analyses for the solutions architect
- Tech: Anthropic Claude (AWS Bedrock), MCP, TypeScript, Node.js, Python, AWS, GCP (Cloud Run, Pub/Sub, BigQuery, Healthcare API), FHIR/HL7, Terragrunt, OpenTofu
Property Insurance GIS Platform — Team Lead (50 developers)
- Led cross-functional team building satellite imagery analysis platform for property valuations
- Architected solution using Kubernetes (AKS), KEDA auto-scaling, Databricks, and Azure Synapse
- Tech: .NET 6/7, Python, Azure Data Factory, Kubernetes, KEDA
Enterprise Banking Platform — Solutions Architect (60 developers)
- Designed and delivered architecture for 10M+ user banking information system
- Responsible for security architecture across distributed microservices
- Tech: Azure, Kubernetes (AKS), KEDA, Big Data pipelines
Spectocor — Healthcare Analytics Recovery — Solutions Architect (over 160-developer program)
- Diagnosed a stalled $15M heart-analytics platform investment; authored a 50-page architectural and program assessment plus an executive brief
- Pivoted from a 160-developer offshore program (5 vendor groups) to a 5-person senior onshore team; personally interviewed and hired all five engineers before handoff
- Project recovered; platform began shipping under the new team
- Tech: .NET, Hadoop, SQL Server, Azure hybrid deployment
Koch Industries — Distributed ML Forecasting & Ranking Engine — Hands-On Architect
- Built a distributed prediction and feature-ranking engine for Koch's internal data-science group, scoring correlations across feature sets up to ~200K features and running hundreds of millions to billions of calculations via massive parallelism
- Integrated an LSTM forecasting module into the prediction engine end to end (design, develop, unit + full integration testing), mastering the PyTorch interface and Ray for parallelism; consulted on DeepAR integration alongside the data-science team
- Evaluated and built distributed-compute POCs (KEDA + Kubernetes on EKS, Ray, Anyscale, Azure Durable Functions) and benchmarked the KEDA solution against the Anyscale prototype; debugged K8s concurrency and database-relationship failures under heavy parallel load
- Bridged the data-science / engineering gap—translated model work into clean input/output contracts and drove unit and integration test strategy for the prediction engine
- Tech: Python, PyTorch, scikit-learn, Ray / Anyscale, AWS EKS, KEDA, SageMaker, RabbitMQ, PostgreSQL
Streaming Platform Architecture — Application Architect (8 developers)
- Led re-architecture initiative for major national streaming provider
- Designed new architecture later adopted by internal engineering team
- Tech: C++, Swift, Objective-C, Java, Android, iOS
Oil & Gas Production Forecasting — Team Lead
- Built Azure Durable Functions application for well output calculations and revenue forecasting
- Tech: Azure Durable Functions, C#, SQL Azure, Event Grid, Go, PostgreSQL
Telecom Cloud Modernization — Cloud Architect
- Led containerization proof-of-concept for major telecom provider's core business services
- Tech: .NET Core, Podman, Docker Swarm, Red Hat Linux, SQL Server, Informix
Software Engineer
- Recruited at age 19 to work on Visual C++ and Visual Basic for Windows CE
- Contributed to Bill Gates' strategic Windows CE initiative
Education
Texas Tech University
Computer Science coursework
Continuous professional development through hands-on enterprise engagements since 1994.
Notable Clients
