Available for engineering opportunities

I build systems that explain why they work.

Software, AI, data, cloud, cybersecurity, DevOps, and game-engine projects backed by architecture decisions, failure tests, performance budgets, and honest limitations.

Explore the work Experience & résumé

evidence.json

focus: ["reliability", "clarity"]
default: "measure before claiming"
failure_mode: "bounded + observable"
delivery: {
  tests: true,
  docs: true,
  rollback: true
}

46: validated projects
22: languages
31: role paths
6: deep dives

Selected case studies

Six systems. Six different engineering risks.

View all 46 projects

01 · Python validated

code-flow-studio

Connects a polyglot editor, isolated compiler workers, bounded fuzzing, revision evidence, and private local accounts.

Read case study Repository

02 · Python validated

business-knowledge-workbench

Combines HTML-first document ingestion, dimension-safe units, provenance, role previews, and encrypted local workspaces.

Read case study Repository

03 · Python validated

personal-image-model-studio

Trains a real caption-conditioned image model with held-out evaluation, rights-aware datasets, and encrypted local accounts.

Read case study Repository

04 · C++ validated

forge2d-engine

Combines a sparse-set ECS, deterministic replays, A*, collision broadphase, fuzzing, and SDL3.

Read case study Repository

05 · Python validated

secure-delivery-pipeline-lab

Verifies signed build identity, SBOMs, tamper resistance, staged promotion, and independently authenticated rollback.

Read case study Repository

06 · Python validated

applicant-evidence-studio

Links job-relevant skills to source evidence, preserves education uncertainty, and offers factual resume coaching without automated hiring.

Read case study Repository

How to evaluate this work

Start with the evidence, not the adjectives.

01
Open a case
Understand the problem, constraints, and design tradeoffs.
02
Inspect the tests
Follow the automated evidence from local checks to CI.
03
Challenge the limits
Read what the project does not claim and what comes next.

Architecture of this site

Static by design. Current by review.

The portfolio does not call GitHub during a visitor request. Repository facts are refreshed into a reviewed cache, case studies are schema-checked, and Astro pre-renders every route.

This keeps the deploy inexpensive, the failure surface small, and the content available even when an external API is not.

Content and repository facts are validated before deployment; visitors receive static HTML and a small precomputed search index.

Looking for a specific fit?

Filter by role, language, or problem domain.

Open the project index