PROJECTS / PRACTICE

Client delivery. Independent products. What comes next.

Systems I have delivered, products I build independently, and directions I am exploring. Each explains the problem, my contribution, and where the work stands today.

Client work is described anonymously. Systems in use, internal tools, and prototypes are labeled separately; design goals are not presented as implemented capabilities.

01 / Client delivery

Enter the business. Build a system people can use.

01 / CLIENT DELIVERY

Investment research & intelligence

Delivered · In research use

Research and intelligence systems for private investment firms, connecting information gathering, evidence organization, AI-assisted analysis, and human review in a traceable research process.

BUSINESS VIEW / CONCEPT SUMMARY

Research task

Sources
Evidence
AI-assisted analysis
Human judgment

A summary of the work’s scope—not a client architecture or automated execution flow.

BUSINESS PROBLEM: Research involves continually interpreting information from multiple sources while separating facts, analysis, and judgment. The system needs to support that process—not just produce an answer that looks complete.
CORE CAPABILITIES: Organize information sources and evidence around research tasks.
Make AI-assisted analysis part of an inspectable research process.
Keep review and final judgment with people.
MY CONTRIBUTION: Led client discovery, research workflow and product design, technical trade-offs, AI-assisted development, testing, delivery, and iteration.
CURRENT STATE: Delivered · In research use
SCOPE & LIMITS: Supports research, not a substitute for investment decisions. No investment-return or autonomous-trading claims.

02 / CLIENT DELIVERY

Supplier fulfillment & quality coordination

In production use · Business-user validation

A coordination system for an export-oriented manufacturer serving European and US customers. It organizes information around orders, suppliers, and quality events to support fulfillment tracking and delivery-risk identification.

BUSINESS VIEW / CONCEPT SUMMARY

Fulfillment coordination

Orders
Suppliers
Quality events
Delivery milestones

A summary of the work’s scope—not a client architecture or automated execution flow.

BUSINESS PROBLEM: Order progress, supplier feedback, and quality issues require coordination across roles. People need a shared view of the business objects they track, the issues they resolve, and the work still outstanding.
CORE CAPABILITIES: Organize order milestones and supplier fulfillment information.
Connect quality issues, rework, and delivery risks.
Provide a shared working view for cross-role coordination.
MY CONTRIBUTION: Led business discovery, business-object and workflow design, system design, AI-assisted development, access-control design, deployment, and delivery validation.
CURRENT STATE: Deployed to production and in use. Client business users participated in validation and confirmed it passed.
SCOPE & LIMITS: Demonstrates business-system delivery and AI-assisted engineering; it does not characterize every business function as autonomous AI.

OTHER DELIVERY / IN USE

Manufacturing operations system

A production-operations information system for a manufacturer, organizing business records and operating information into a queryable, collaborative working view. I led requirements discovery, product design, and engineering delivery. The system is live and in use, with client business users involved in validation.

These cases do not disclose client names, internal data, or unauthorized system interfaces.

02 / Independent products & open tools

Make enduring questions into inspectable products and tools.

SignalOS

A research system built around evidence, industry analysis, and human judgment, exploring how AI-assisted analysis can become part of a traceable workflow.

Explore project

Golf Meridian

A WeChat viewing companion that brings together tournament calendars, course information, and AI-assisted, human-reviewed event interpretation.

Explore project

CollaborationOS

A governance framework for consequential human–AI collaboration, keeping context, evidence, authority, and accountability inspectable.

Explore project

NativeHelix

AI-native product and architecture methods organized into reusable Skills and Capability assets.

Explore project

Kairos

An observation and reading system built around macro cycles, research evidence, and open questions.

Explore project

The anonymous investment-research work and SignalOS reflect related practice. The case describes client delivery; the product page explains system design. They are not counted as separate outcomes.

Explore public work on GitHub

03 / Research & prototypes

Test the next capability through design, prototypes, and real tasks.

03 / INTERNAL CAPABILITY

Domain knowledge & expert agents

In internal use · Ongoing development

An internal knowledge platform that organizes domain knowledge, source evidence, and task methods, with domain-expert agents supporting project analysis, solution design, and delivery preparation.

BUSINESS VIEW / CONCEPT SUMMARY

Domain-expert capability

Knowledge structure
Source governance
Expert research
Task methods

A summary of the work’s scope—not a client architecture or automated execution flow.

DOMAIN COVERAGE: Coverage includes retail and consumer businesses, manufacturing and supply chains, securities and capital markets, with related practices in growth, commerce operations, financial risk, and AI engineering.
CORE WORK: Knowledge architecture: structure domain concepts, metrics, methods, and task specifications.
Source and quality governance: select material, review content, and retain traceable evidence.
Expert research: use interviews to deepen understanding of business contexts and practical constraints.
Expert-agent design: define roles, task context, collaboration boundaries, and human-review requirements.
MY CONTRIBUTION: Independently designed the knowledge structures, selected sources, reviewed content quality, and conducted expert research; led development of the knowledge platform and domain-expert agents.
CURRENT STATE: The knowledge platform is in internal use; expert capabilities continue to be developed and refined.
SCOPE & LIMITS: Supports professional analysis; it does not replace domain experts or accountable decision-makers.

IN DEVELOPMENT / PRODUCT & ARCHITECTURE DESIGN

Ontology-driven business applications

I am designing an ontology-driven application system built around business objects, relationships, rules, and approved actions—exploring how enterprise data, domain knowledge, AI analysis, and business operations can work together.

CONCEPTUAL DESIGN / NOT A LIVE ARCHITECTURE

Business context

Objects
Relationships
Rules
Approved actions

Design concepts and their shared business context—not a runtime data flow.

WHAT I AM DESIGNING

Objects & relationships: Define business entities and how they relate.
Rules & context: Organize task constraints, business meaning, and the basis for judgment.
Knowledge & capabilities: Design how domain knowledge, AI analysis, and deterministic tools work together.
Action & human authority: Define which operations may be proposed, which need approval, and who retains responsibility.

MY CONTRIBUTION: Leading product definition and overall architecture, including the connections between data foundations, expert capabilities, and client-side operating environments.
CURRENT STATE: Product definition and overall architecture design; preparing to begin development. Not yet live.

These are design directions, not a released platform or validated client capability.

PROTOTYPE / NOT LIVE

Commerce operations decisions

Solution design and prototype POCs around commerce analysis, option comparison, and human approval, exploring how operating capabilities could fit into merchants’ working environments. Not deployed or live.

PROTOTYPE / NOT LIVE

Group finance scenarios

Requirements analysis, solution design, and prototype validation for group finance, exploring how AI-assisted analysis can support human decisions. Not deployed or live.

Explore the FDE approach

DISCUSS A SIMILAR PROBLEM

Start with the business context.

You do not need to choose a model or agent first. Begin with who does the work, where it gets stuck, and the available data and constraints. Then we can discuss what is worth building and testing.