Denver, Colorado
Senior Software Engineer | Agentic AI Systems
Python · LangGraph · FastAPI · Neo4j · TypeScript · React
Senior software engineer with about 20 years of professional and independent development experience across data-intensive applications, graph visualization, enterprise authentication, developer tooling, testing, and Artificial Intelligence systems.
Professional summary
Joshua designs and implements systems across the full stack — Python and FastAPI services, React and TypeScript interfaces — spanning data, evaluation, and infrastructure boundaries. Current independent work applies Python, LangGraph, FastAPI, Neo4j, structured data contracts, retrieval, tool integration, and replay evaluation to agentic Artificial Intelligence (AI) systems, with emphasis on the reliability, observability, and evaluation that let teams trust what their agents do in production. The target is senior software engineering for AI products and platforms, not machine-learning research.
Selected engineering evidence
Three examples spanning production performance, complex visualization, and current agent evaluation work.
Reduced processing time for a production CSV import handling more than 5,000 rows under real customer load by removing N+1 query patterns, batching database operations, memoizing repeated lookups, and hardening edge-case parsing.
Built Vue, TypeScript, D3.js, and Apache ECharts interfaces for relational datasets, with dynamic filtering, styling, grouping, and performance work for large network graphs.
Built a typed LangGraph pipeline with structured Pydantic validation, weighted model-assisted scoring, policy filtering, versioned prompts, and replay fixtures that produce per-node audit traces plus reliability, cost, and latency metrics per run — for observability, governance, and reliability regression review.
Flagship projects
These are active engineering projects. Status and limitations are stated directly.
Local-first evidence-graph tooling that turns documents, configuration, source anchors, and Git history into source-traceable artifacts, giving agent retrieval an auditing, visibility, and governance layer so teams can see and trust what agents pull in.
Current limits: early-stage, source-install software with a command-line interface; no hosted service or local user interface. Semantic retrieval is experimental and off by default.
A browser space shooter built with React, TypeScript, and a full automated test suite, designed to teach Vim commands through keyboard-driven gameplay.
Current limits: the game remains in active development; its tutorial covers a deliberate subset of Vim commands, and no adoption or learning-effectiveness claims are made.
Core technical capabilities
Python, LangGraph, LangChain Core, OpenAI API, tool calling, structured outputs, retrieval-augmented generation, graph-backed retrieval, agent evaluation harnesses, reliability, cost, and latency metrics, policy and governance controls, workflow orchestration, prompt and context engineering, Model Context Protocol integration, and AI developer tools (OpenHands, Claude Code, Cursor, Copilot).
FastAPI, PHP, REST APIs, Neo4j, PostgreSQL, MySQL, Redis, Structured Query Language, knowledge-graph and relational modeling, caching, and data-pipeline optimization.
React, TypeScript, Vue, JavaScript, D3.js, Apache ECharts, Phaser, Three.js, responsive interfaces, network visualization, and Geographic Information System interfaces.
System design, performance analysis, unit, integration, and end-to-end testing, replay and regression evaluation, observability, agent evaluation, GitHub Actions, continuous integration and delivery, Docker, and Google Cloud Platform.
Professional experience
Independent Software Engineer - Agentic AI Systems
Denver, Colorado
- Present
Current flagship work includes EvidEngine and VimCommander, described above, supported by focused prototypes and evaluation workflows for agentic systems.
Software Engineer, Full Stack
Denver, Colorado
-
Full-stack product engineering for data-heavy network-analysis applications. Performance and graph-interface results are documented in Selected Engineering Evidence above.
Earlier experience
Drupal Developer
Manchester, New Hampshire
-
Full-Stack Developer
- , intermittent
Contact and résumé download
For roles spanning agent architecture, evaluation, developer tooling, full-stack product engineering, data-intensive systems, and technical integration, contact Joshua directly.