01
TCAF
Shows how I structure governance-first, spec-driven AI delivery around shared boundaries, repository evidence, developer review and explicit verification.
Projects
Public work that shows how I structure software projects.
These projects are not isolated code samples. They show product decisions, scope, planning, implementation, verification and delivery choices that can be inspected directly.
Public release · v0.4.0
Task-Contract AI Development Framework — TCAF
A governance-first, spec-driven framework for controlled AI software delivery.
TCAF implements Task-Contract Development, bringing Spec-Driven Development, Agentic / Harness Engineering and Software Delivery Governance into one operating model. Repository evidence defines the current state, Task Contracts bound what AI may change, and verification plus developer review remain explicit before work is accepted.
The objective is simple: gain useful AI acceleration without losing scope control, project context, working code or human technical responsibility.
Spec-driven task definition
Task Contracts and explicit boundaries
Repository evidence as source of truth
Preservation-first rules
Verification and developer review
Versioned runtime and documented workflows
Public project · work in progress
Software Delivery Planner
A lightweight application for planning and monitoring software delivery in one place.
It connects project scope, features, estimates, risks, KPIs, architecture and delivery status instead of leaving planning material scattered across separate documents and tools.
Project and feature flows
Scope, estimates and risks
Shared frontend/backend contracts
Automated API and UI tests
OpenAPI / Swagger documentation
Architecture and backlog material
Public project · MVP complete
Travel AI Flight Optimizer
Compare flights across multiple dates, plan the transport around them and get useful destination information in one flow.
The product combines live flight discovery, flexible-date and multi-airport comparison, transport planning to and from airports, source-backed destination research and exportable trip reports. AI is used for selected research and synthesis tasks, while ranking and compatibility decisions remain deterministic.
Live flight search through Kiwi MCP
Exact and flexible date comparison
Multiple departure airports
Home-to-airport transport planning
Source-backed destination insights
Deterministic shortlist ranking
Standalone trip reports
React + TypeScript frontend, JavaScript/Express backend
What these projects show
Product, delivery and implementation in the same portfolio.
Each project highlights a different part of the work, but together they show the same operating model: define the problem, structure the work, make decisions visible and keep implementation verifiable.
02
Software Delivery Planner
Shows how I translate project-management information into a working software product and visible delivery model.
03
Travel AI Flight Optimizer
Shows end-to-end product work: user need, provider integration, deterministic logic, AI-assisted research and a complete usable flow.
Inspect the work
Repositories and documentation are public.
You can inspect implementation, documentation, current status and known limitations directly, rather than relying only on a portfolio description.