status: open to Principal / Staff AI engineering roles

John William Baker

Principal full-stack engineer building LLM agents, RAG systems and the dashboards that tell you whether they actually work.

14+ years shipping enterprise software at Oracle, Sun Microsystems and Lawrence Livermore National Laboratory. At Sun they called me "Dashboy" for the dashboards. These days the dashboards trace agents.

See the work LinkedIn GitHub

Years shipping

14+

Years at Oracle HCM

14

Customers on my AI agent

600

Live apps

2

Surging on iOS · AI Wire

trace: career

Experience

Oracle HCM Cloud

Principal Applications Developer · Benefits & Talent

duration: 14 yrs

Built Java REST APIs and JavaScript front ends for HCM Benefits, then moved into agentic AI. Developed the customer-facing HCM Calculator agent, an LLM that asks guiding questions to help employees allocate benefit amounts without reading or submitting personal data, launched to 600 customers.

Sun Microsystems

Software Engineer · Dashboards

alias: "Dashboy"

Built the internal dashboards that earned the nickname, and the habit of making system behavior visible that still drives how I build today.

Lawrence Livermore National Laboratory

Software Engineer · Federal research lab

env: restricted

Engineering inside a locked-down federal environment, where approved tooling and careful data handling were the default, not an afterthought.

education: B.S. Computer Science, California State University, Chico

GET /projects

Selected projects

Agents, retrieval and the observability around them, built from scratch and open on GitHub.

RAG · AGENTS · EVALS

Benefits Support Triage Agent

Dual-corpus RAG over policies and resolved tickets with pgvector, a 4-tool MCP server, a bounded agent loop with token budgeting and OpenTelemetry tracing, and a 40-case golden eval suite gated in CI.

View repo →

LANGGRAPH · HITL

Life-Event Agent

An HR life-event case agent in LangGraph with human-in-the-loop approval, Postgres checkpointing and evals, the workflow I lived for years in HCM Benefits, rebuilt agent-first.

On GitHub →

LIVE · LLM FEED

AI Wire

A live AI-industry news feed that refreshes while open, with LLM-classified filters and a swappable model provider behind it.

Open AI Wire →

OTEL · SLOs

Agent Latency Lab

OTLP tracing with W3C traceparent propagation and Monte Carlo SLO simulation, published as a VS Code Marketplace extension with a full GitHub Actions release pipeline.

On GitHub →

INTEGRATION · POLICY

NexusAgent

An enterprise AI integration hub: tool registry, policy engine, JSONL audit log and SSE streaming, running on the Anthropic API or local Ollama models.

On GitHub →

iOS · SHIPPED

Surging

A goal-tracking app on the App Store, built with React, TypeScript and Capacitor and coached by Groq-hosted Llama models.

App Store →

cat stack.json

Stack

AI / LLM

Anthropic API Ollama LangGraph MCP pgvector DeepEval

Application

React TypeScript Node.js Python Java

Data & Ops

PostgreSQL Redis OpenTelemetry GitHub Actions

Building agents that need to hold up in production?

Based in the East Bay, California. Happy to talk roles, architecture or evals.

johnbakerjobs@gmail.com linkedin.com/in/johnwilliambaker github.com/juubaker