Executive technical portfolio · 40+ years of programming, software, web development and production problem-solving
AI knowledge systems · vector databases · production engineering

I turn complex information into working AI systems.

I am James Murray, a programmer, web developer, software designer and technical builder with more than four decades of hands-on experience—from early software development and large-scale web publishing to secure APIs, private vector databases and production AI search.

This page brings together the technical architecture, business thinking, recovery-industry knowledge and 94 working portfolio projects that demonstrate how I take an idea from raw data to a live, secure and recoverable production system.

40+Years programming
94Portfolio projects
6,506Therapist profiles
26,725Vector knowledge chunks
100/100API regression tests
Executive profile

A builder who understands the entire system—not just one layer.

My strength is connecting strategy, data, programming, infrastructure, security and user experience. I can investigate an unfamiliar problem, design the architecture, write the processing tools, structure the data, deploy the services, integrate the website, test the result and document how to recover it.

Long-term technical depth

More than forty years spanning software design, programming, databases, websites, hosting, automation, search, media systems and emerging AI infrastructure.

Production execution

I move beyond concepts and prototypes into live systems with service management, authentication, rate limiting, logging, health checks, backups and rollback plans.

Business and human context

Deep experience in addiction treatment, recovery information, provider marketing and the real questions families ask before contacting a therapist or treatment centre.

Career evolution

Four decades of adapting to every major shift in computing.

The common thread across my career is learning new systems quickly and using them to solve practical problems.

Foundation

Programming and software design

Built software and data-driven systems long before modern web frameworks, developing the logic, persistence and troubleshooting discipline that still underpins reliable engineering.

Web era

Large-scale web development and publishing

Created and managed extensive website networks, PHP applications, MySQL databases, content systems, search pages, SEO structures, media libraries and hosting environments.

Automation

Python, PowerShell and structured data pipelines

Built tools to discover websites, crawl difficult sources, clean HTML, reconcile records and generate CSV, JSON, JSONL and SQL imports at scale.

AI future

Private vector databases and grounded answer systems

Designed multi-server AI knowledge platforms using embeddings, Weaviate, FastAPI, OpenAI models, secure PHP proxies and production-grade retrieval workflows.

Featured production system

AddictionTube private AI knowledge and vector-search platform

A complete architecture that transforms therapist and treatment-provider information into searchable, source-grounded answers across a Windows development workstation, Linux VPS and public hosting server.

1

Local data engineering

PowerShell, Python virtual environments, website discovery, crawling, content extraction, deduplication, audits and structured exports.

2

Vector knowledge layer

Normalized profile records, stable identifiers, metadata, chunk types, embeddings, controlled imports and Weaviate collections.

3

Secure VPS APIs

FastAPI, Uvicorn, systemd, semantic search, city browsing, grounded questions, logging, staging, health checks and rollback.

4

Public website integration

PHP proxy, cURL, JavaScript UI, environment secrets, encrypted image tokens, request validation and rate limiting.

AddictionTube private vector database system flowchart
What was built

From websites and spreadsheets to a live AI search experience

The system does not simply send a question to a general chatbot. It searches a private, structured knowledge base built from provider-specific information, applies location and identity filters, removes duplicate chunks and answers from the stored evidence.

Data pipeline: discovery → crawl → extract → clean → normalize → audit → JSONL/CSV/SQL generation.
Knowledge model: identity, practice details, concerns, approaches, contact details, locations, source URLs and profile descriptions.
Retrieval: semantic vector search, exact location filters, unique-profile collapse and provider-specific question answering.
Production controls: shared-secret authentication, HMAC visitor tokens, AES-GCM image tokens, rate limits, structured logging and rollback services.
Technical capabilities

The tools required to take AI from concept to production.

This is an end-to-end skill set covering the parts that usually require several separate specialists.

Data engineering

Website discovery, crawling, HTML extraction, normalization, deduplication, record matching, audits, CSV, JSON, JSONL, SQL and lookup generation.

AI and vector search

Embeddings, RAG, Weaviate, metadata filtering, chunk design, semantic retrieval, grounded answers, fallback logic and source traceability.

Python engineering

Command-line tools, virtual environments, batch processing, API clients, validation scripts, test runners, structured logging and recovery utilities.

VPS and Linux operations

SSH, remote administration, systemd, Uvicorn, service ports, environment files, journal logs, permissions, health checks, backups and rollback.

PHP and website integration

Public interfaces, secure server-side proxies, cURL, JavaScript, forms, pagination, protected image delivery and hosting configuration.

Security and reliability

Secret management, header authentication, HMAC, AES-GCM, rate limiting, privacy-safe logs, checksums, staging promotion and reconstruction documentation.

PythonPowerShellPHPJavaScriptHTML/CSSSQLMySQL/MariaDBFastAPIUvicornWeaviateOpenAI APIREST APIsLinuxSSHsystemdcURLJSONJSONLCSVHMACAES-GCMGit-style backup discipline
Complete work archive

94 projects across AI, recovery, software, business and human systems.

Every card links to a full project page. Use the search or category filters to explore the breadth of the work.

Showing 94 projects
Why this matters to employers and clients

I can own the whole problem and get it running.

What a technical leader will recognize

  • Architecture across local development, Linux infrastructure and public hosting.
  • Practical data-engineering experience with inconsistent real-world sources.
  • Production vector-search implementation rather than a laboratory demo.
  • Security, testing, observability, cost controls, backup and rollback discipline.
  • The ability to write code while also understanding users, content and business outcomes.

Engagements I am suited for

  • AI knowledge-system architecture and implementation.
  • Website-to-vector-database conversion.
  • Python data pipelines, crawling and structured-data modernization.
  • Private semantic search and grounded question-answering systems.
  • Technical audits, rescue projects and production stabilization.
  • Consulting, contract development, fractional technical leadership and special projects.
Consulting · contract · technical leadership

Bring me the difficult project that crosses data, AI, servers and the web.

I am based in Kelowna, British Columbia, Canada, and available for consulting, contract development, architecture, data engineering, vector-database implementation, API development, production integration and technical rescue work.