Projects

AI systems, case studies, and reusable product patterns.

A focused view of the work I want visitors to inspect: agent workflows, RAG trust systems, evaluation dashboards, and project-based AI teaching systems.

Start with flagship case study View resources

Start from the user workflow, not the model.

Make retrieval, tool use, and evaluation visible.

Design for failure modes before launch.

Connect frontend states to backend decisions.

Flagship case study

The project visitors should inspect first

Flagship case study

Agentic AI Production Harness

A full-stack case study for turning an AI agent demo into a controlled, observable, and testable product workflow.

Outcome

Demonstrates how agentic AI can move beyond prompt demos into workflows with tool boundaries, traces, approval gates, and evaluation loops.

Architecture

Workflow UI, backend orchestration, retrieval, tool gateway, evaluation runner, trace dashboard, and release-readiness checks.

Next.jsFastAPIPostgreSQLpgvectorRedisDockerLLM APIs

Open related artifact

System patterns

Reusable architectures behind my AI work

RAG system pattern

RAG Knowledge Assistant System

A product architecture for internal knowledge assistants where users need sourced, reviewable, and correctable answers.

Outcome

Shows how retrieval quality, citations, metadata, permissions, and feedback loops become product trust features.

Architecture

Document ingestion, metadata enrichment, vector retrieval, answer generation, citations, feedback capture, and admin review.

ReactNode.jsPythonPostgreSQLVector searchOpenAI

Open related artifact

Evaluation system pattern

LLM Evaluation Console

A dashboard pattern for comparing prompts, reviewing outputs, tracking traces, and catching regressions before release.

Outcome

Shows that production AI quality depends on repeatable measurement, not one-off manual testing.

Architecture

Evaluation datasets, rubric scoring, run history, model comparison, trace review, regression checks, and release signals.

Next.jsPythonPrismaDashboardsEval rubricsCI checks

Open related artifact

Teaching system

AI Workshop Delivery System

A teaching system for converting AI engineering topics into hands-on projects, code walkthroughs, and structured learning paths.

Outcome

Shows how complex AI engineering concepts can be taught through projects, diagrams, guided exercises, and feedback loops.

Architecture

Learning paths, project templates, student notebooks, code walkthroughs, workshop pages, practical assignments, and feedback loops.

Next.jsJupyterPythonMarkdownDiagramsGitHub

Open related artifact

What this page demonstrates

My project work focuses on the parts of AI engineering that matter after the demo: workflow design, data trust, bounded tool use, evaluation, observability, and teaching clarity.

Work with me Read writing