Project library

Applied AI implementations, engineering systems, and delivery proofs.

The library covers AI-enabled applications, backend and automation systems, reporting workflows, and ML delivery. Each project is grouped by maturity and evidence, with public links only where intentionally ready.

Featured

Public repos

Private

Homepage showcase

Start here

Three strongest Applied AI and engineering entry points.

These pages show the strongest mix of visual evidence, implementation depth, runnable proof, and repository readiness.

AI Nutritionist

Benchmark project: generated meal plans, Hybrid Recommender V2, weekly planner summaries, data/model cards, Dockerfile healthcheck, CI, screenshots, and 95 pytest tests.

OpsForge Suite

Integrated platform: planning, review queue, reporting, audit, exports, verified Docker Compose smoke, seeded data, and screenshots.

OpsForge Planning Workspace

Best planning workflow: rough operational briefs become dossiers, saved runs, compare notes, and export manifests.

Flagship

Applied AI and product-grade engineering systems

The strongest project set combines AI-enabled capabilities with polished application workflows, runnable evidence, tests, screenshots, and clear repository status.

Neural nutrition recommender

AI Nutritionist

Public

Local nutrition planner that generates daily and weekly meal plans from USDA/FNDDS data, a Mediterranean/Greek extension, a reviewed recipe-backed pilot, Hybrid Recommender V2 ranking, bounded weight-loss targets, local feedback, flat and ingredient grocery exports, Streamlit, CLI, FastAPI, and 95 pytest tests.

Impact

Generates daily and weekly meal plans with profile targets, weekly planner summaries, bounded weight-loss targets, diet-mode guardrails, alternatives, flat grocery export, recipe ingredient export, and local feedback.

Public RepoML RecommenderFastAPI95 Tests

Validation

95 pytest tests

Ruff + mypy

GitHub Actions CI

PythonStreamlitFastAPIPandasscikit-learn

Case study Repo ZIP

Operations software platform

OpsForge Suite

Private

Private local-demo platform that combines planning workspace, operations review queue, reporting center, scoped memory/cache, audit trail, and export handover flows in a production-shaped Docker Compose stack.

Impact

Turns multiple smaller portfolio ideas into one coherent operations platform with planning, review, reporting, audit, and handover modules.

Private RepoDocker SmokeFastAPI + Next.jsPostgreSQLSeeded DemoExport Packages

Validation

13 backend tests

Docker Compose smoke

API contract checks

Next.jsFastAPIPostgreSQLDocker ComposeSQLAlchemy

Case study

CaseForge Studio local web app showing a generated implementation blueprint

Operational implementation planning

OpsForge Planning Workspace

Public

Public planning workspace that turns operational software briefs into implementation dossiers, architecture notes, risk registers, test plans, saved-run comparisons, and export manifests.

Impact

Turns rough operational implementation briefs into handover-ready planning artifacts before build work starts.

Public RepoPlanning WorkspaceDeterministic CoreDocker

Validation

24 unittest tests

Package build

OpsForge module contract test

PythonCLIHTTP APIlocal web UIunittest

Case study Repo ZIP

Job Market Intelligence Pipeline report desktop screenshot

Repeatable reporting system

OpsForge Market Intelligence

Public

Public reporting system for fixture-backed source collection, validation, DuckDB history, source quality, run-over-run deltas, report index, and executive summaries.

Impact

Turns fixture-backed source inputs into a repeatable intelligence report with source quality, history, run-over-run movement, and a reviewer-ready report index.

Public RepoReporting SystemDuckDB HistoryCI

Validation

28 pytest tests

OpsForge module contract test

Two-run fixture demo

PythonTyperDuckDBPandasBeautifulSoup

Case study Repo ZIP

Systems

Workflow and reporting proof

Focused public systems that demonstrate operational intake, human review, deterministic reporting, and repeatable artifacts without pretending to be large platforms.

AI Workflow Automation Assistant home screen with provider and review boundaries

Operations intake and human review

OpsForge Review Queue

Public

Public operations intake review queue for synthetic commissioning, configuration, test-failure, handover, SQL/data-validation, and reporting-anomaly requests, now with reviewer action packages.

Impact

Turns operational intake into a review queue with validation, classification, priority, provider boundaries, action packages, persistence, seeded demo states, status history, and human approval.

Public RepoReview QueueHuman DecisionDocker

Validation

30 pytest tests

ruff

Action-package API/UI checks

FastAPIPydanticSQLAlchemySQLiteOpenAI optional

Case study Repo ZIP

Marketing-ops control report

LinkedIn Ads Control Report

Public

Daily marketing-ops report that turns LinkedIn Ads exports into KPI health, pacing verdicts, daily movement, control risks, campaign movement, and HTML/JSON/Markdown artifacts.

Impact

Turns daily ad CSV exports into KPI cards, pacing signals, day-over-day movement, ranked review actions, and a compact risk register.

Public RepoReportingDeterministic KPIsHTML Outputs

Validation

17 unittest tests

OpsForge module contract test

One-command report refresh

PythonCSVHTML reportsJSONMarkdown

Case study Repo ZIP

Supporting

ML delivery support

Supporting public work that stays honest about model artifacts, demo boundaries, and what is needed before promotion to a flagship project.

ViT Pet Classification Pipeline Streamlit UI without third-party branding

Computer vision release boundary

ViT ML Delivery Proof

Public

Supporting ML-delivery project with Oxford-IIIT data preparation, ViT fine-tuning metadata, eval JSON, confusion matrix output, release checksum manifest, FastAPI/CLI/Streamlit surfaces, and explicit artifact boundaries.

Impact

Shows ML delivery packaging beyond notebooks: public dataset preparation, training metadata, eval JSON, confusion matrix output, release checksums, CLI, API, Streamlit UI, and artifact boundaries.

Public RepoML DeliveryEval ArtifactsFastAPIStreamlit

Validation

13 pytest tests

ruff

Oxford-IIIT subset smoke

PyTorchTransformersHugging FaceFastAPIStreamlit

Case study Repo ZIP

Private research

Public-safe research case study

Private research shown only at the system-design level. Source links, raw outputs, credentials, and unsafe operational details are intentionally hidden.

Paper-only market replay research lab

Agentic Trading Research Lab

Private

Private paper-only research platform for volatile market-event study, event-harsh replay, deterministic risk gates, offline simulation, audit logs, and dashboard/report artifacts.

Impact

Shows how to design a risk-gated research system without exposing private trading logic or making live-trading claims.

Paper ResearchNo Live TradingRisk-GatedActive Development

Validation

pytest suite

paper-only safety audit

event-harsh replay checks

PythonCLIpytestDockerSQLite

Case study