JJ Salley
AI Systems Developer · Machine Learning Engineer · Decentralized Systems
Software engineer and AI/ML consultant with 10+ years of industry experience. Former Software Design Engineer in Test at Microsoft (Visual Studio, MSDN, MSN, Business Intelligence). Currently building AI-driven solutions, LLM integrations, agentic infrastructure, and security tooling for clients across industry and academia.
I design systems with correctness and failure modes in mind — whether that's sandboxed LLM execution, security scanning pipelines, or evaluation frameworks for graduate-level AI curriculum.
🔧 What I Build
- Agentic Infrastructure — LLM runtimes with sandboxed execution, command whitelisting, audit logging, and semantic search via embeddings
- MCP Servers — Early adopter of Model Context Protocol; built practical tool-use servers in Go before widespread industry adoption
- Security Tooling — Portfolio-wide vulnerability scanning, SBOM generation, license compliance automation
- AI/ML Systems — Vector database interfaces, LLM-powered document processing, data pipelines, and ML model evaluation
- Collaborative Systems — Real-time editors using CRDTs (Yjs, Automerge), presence/awareness packages
- Developer Tooling — CLI tools, Chrome extensions, Vim plugins, cross-platform mobile apps
🏗️ Featured Projects
| Project | Description |
|---|---|
| llm-runtime | Secure LLM File Access Tool — models autonomously explore repositories via Docker-sandboxed execution with command whitelisting, path validation, and audit logging |
| mcp-mood-quotes | MCP server in Go — early demonstration of the Model Context Protocol tool-use standard |
| vectordb-interface | Go interface layer for vector database operations, enabling semantic search and embedding-based retrieval |
| vulnsweep | Portfolio-wide security scanner built on Trivy — automates SBOM generation, vulnerability checks, and license compliance across repositories |
| SiteVigil | Uptime monitoring tool — continuously checks health of managed websites and services, alerting on failures |
| collab-editor suite | Multiple real-time collaborative editor implementations using Yjs, Automerge, TipTap, and Node.js |
💼 Experience
- Microsoft — Software Design Engineer in Test (7.5 years). Designed test plans, defined pass/fail criteria, wrote test automation, reviewed specifications, managed testers. Product groups: Visual Studio, MSDN, MSN, Release, Business Intelligence Applications.
- Bellevue Business Solutions — AI/ML consulting, LLM integrations, MCP server development, security automation, and data-driven solutions for clients across industry and academia.
- Subject Matter Expert — Curriculum congruency evaluation for graduate-level ML, AI, and Architecture courses at a national university. Rubric design, gap analysis, and quality evaluation for technical content.
- Data Analytics Tutor — Graduate and undergraduate instruction in ML fundamentals, model evaluation, Python (Pandas, NumPy, Scikit-learn), SQL, and statistical methods.
🎓 Education
- MS, Data Analytics — Western Governors University
- MBA, Information Technology Management — Western Governors University
- MS, Computer Science — University of Texas at Tyler
- BA, Interdisciplinary Studies & BA, Sociology — University of Missouri-Columbia
📄 Certifications
Microsoft Certified: Azure AI Fundamentals · CompTIA A+ · CompTIA Security+ ce · IT Project Management · Google Digital Marketing & E-Commerce · Certified Professional Coach · 10+ FEMA certifications
💬 Areas of Expertise
- Agentic workflows and LLM tool-use patterns (MCP)
- AI/ML evaluation, benchmark design, and rubric development
- Security automation and DevSecOps
- Data science with Python (Pandas, Scikit-learn, NumPy, Matplotlib)
- Test design, quality evaluation, and failure mode analysis
- Real-time collaboration systems (CRDTs, Yjs, Automerge)
📫 Contact
- Email: jjsalley@gmail.com
- LinkedIn: JJ Salley