Product leader × hands on builder

I turn frontier tech into products that ship.

I lead crypto products and stay close to the work. These projects show how I turn new technology, especially AI agents and systems on chain, into tools people can actually use.

Explore selected work GitHub

Public workReal, open source contributions across Injective products and developer tools.

AI agentsIdentity, reputation and tools that let agents interact with real products.

Built end to endFrom the original product idea through implementation, testing and launch.

SHIPwith evidence

Product

Agents

Protocol

Execution

I connect product thinking, technical decisions and execution, then verify that the result works.

Selected systems / 2026

Selected work

Systems, not demos.

The public Injective projects show how I turn emerging technology into products developers can actually use. The independent products show the same approach applied end to end, from product idea to a working system.

Public · Shipped01

Injective Agent Identity Stack

Developer tools that give AI agents a verifiable identity on Injective. Builders can register an agent, attach its wallet and public profile, and make its services and reputation discoverable.

Role: Defined the product direction and worked directly on the implementation

SDK + CLIRegister and manage agents from code or a terminal

Agent cardsA public profile explaining what each agent does

Wallet linksProof that a wallet belongs to the agent

DiscoveryFind agents and review their track record

ERC-8004TypeScriptEIP-712IPFSDX

Public · Under review02

Injective MCP Tools for AI Agents

A set of tools that lets AI assistants create, find and manage Injective agent identities directly from a conversation or automated workflow. It connects the identity system to Claude and other MCP compatible clients.

Contribution: Designed and built the complete identity and reputation integration

Tools added8

Tests passing on the branch321

MCPAgent SDKReputation

Public · Shipped03

Launching New Helix Markets

Launch work that made new USDC grid trading markets available in Helix. I corrected market names, linked each market to the right contract on chain, and updated the files used by production and staging.

Role: Product lead working directly in the launch path

Add markets→Link contracts→Verify data→Launch

HelixGrid tradingUSDC

Public · Personal04

Agentic Delivery Playbook

A practical playbook for using AI agents to deliver real product and engineering work safely. It helps teams choose the right level of control, verify the result and keep a clear record of what changed.

Role: Creator

Assess task→Choose controls→Verify work→Keep evidence

AI deliveryCodexControls

Live on Google Play · Private source05

SafeReceipt

A privacy focused Android app that keeps important receipts available for warranties, returns, expenses and insurance claims. Users can scan paper receipts or import PDFs, extract and search the important details, organize receipts, export them and back them up to their own Google Drive.

Role: Built and launched the product end to end, including mobile UX, OCR, local storage, backup, monetization and the Google Play release.

Scan or import→Extract details→Organize→Find later→Export or back up

AndroidReact NativeOCRLocal firstGoogle Drive

View on Google Play

Private · Access shared06

Knowledge Bits

An AI content production system that turns a brief and trusted sources into a complete package ready for review: research, written content, media, quality checks and final delivery. Every source and approval remains traceable.

Role: Designed the product, architecture and implementation

Research→Draft→Check→Create media→Approve→Deliver

EvidenceHuman reviewWorkersPostgres

Private · Access on request07

Farming, Guarded Trading Copilot

A trading copilot that first observes and proposes actions, then requires human approval before it can execute. It is designed to make automation useful without giving an agent unchecked control over funds.

Role: Designed the product, safety model and implementation

Analyze→Propose→Approve→Execute→Record

TradingShadow modeApprovalsSafety

Operating style

Strategy close to execution.

I am most effective when product judgment, technical fluency and delivery ownership live in the same loop. The goal is not to write more artifacts. It is to make better decisions and turn them into working systems.

Make the product idea concrete

Turn new technology into a clear user problem, a useful outcome, and a simple measure of success.

Build one complete path

Make the full user journey work in reality before polishing disconnected pieces.

Prove that it works

Use tests, visible system state, and human approval where the consequences are meaningful.

01 / FRAMEClarifyWho it helps, what changes, and what it may do.

02 / BUILDIntegrateBuild the smallest complete user journey.

03 / PROVEValidateTest the happy path and how it fails.

04 / SHIPOperateShip, learn, and improve the product.

Continue the conversation

I’m at my best where product judgment and technical execution meet.

The public repositories and SafeReceipt app are linked above. I can grant access to the source for SafeReceipt, Knowledge Bits and Farming for a closer technical review.

Email Joan Save as PDF