WENQING JI / 2026

SOFTWARE DEVELOPMENT · AI AGENTS · BACKEND SYSTEMS

Hi, I'm Wenqing.I build reliable software for uncertain systems.

Master of IT · First Class Honours

Auckland, New Zealand · Open work visa

02 / OVERVIEW

A practical engineering thread

I'm an Auckland-based software developer. I completed a Master of Information Technology with First Class Honours at the University of Auckland in September 2026, following an undergraduate degree in information systems.

I build Python services, React and Next.js interfaces, and AI-agent automation. My independent Windows runtime is released as Execution 0.1.8; my event-driven Java project explores accounting, quoting, and risk controls.

During my University of Auckland Industry Internship, I contributed reusable frontend components, REST API integration, prompt and search logic, testing, and documentation in a team setting.

I'm looking for graduate and junior opportunities across New Zealand where I can contribute practical software, learn from experienced engineers, and help make systems easier to test and diagnose.

03 / SELECTED WORK

Evidence before claims

A released Windows execution runtime for MCP-compatible agents, with visual grounding, evidence-bound actions, result verification, and confirmed updates with rollback.

Stack

Runtime & interface

Python
FastAPI
MCP

Visual grounding & review

Computer vision
OpenAI API

Verification & observability

pytest
Structured receipts
Diagnostic logging

Evidence

Execution 0.1.8 is released with exact-target update confirmation, task-aware maintenance locks and rollback.
Current-agent mode completed 20 live 7-Zip steps; installed 0.1.7 separately completed a reviewed Calculator workflow.
The 0.1.8 development regression group recorded 502 passes and two skips; overlapping follow-up checks are reported separately.

Limitation

Current-agent mode supports basic clicks only. Wider API-route plans have separate acceptance; general unattended reliability is not established. Learning is paused.

View case study →

A five-person University of Auckland CS778 team prototype that explored a staged AI conversation and confirmation workflow.

Stack

Frontend

Next.js
React
TypeScript
Tailwind CSS

Frontend verification

Jest
Playwright

System context

FastAPI
PostgreSQL
AWS Bedrock

Evidence

The coursework prototype was delivered by a five-person University of Auckland CS778 team.
Frontend work covered the project workspace, staged conversation interface, summary-confirmation flow, and report presentation.
Jest and Playwright were part of the frontend verification stack.

Limitation

This is the original coursework prototype, not the current independently operated IdeaSense product.

View case study →

An event-driven trading-system challenge with Taker, Quoter, and Hedger services, risk controls, and ledger reconciliation.

Stack

Services

Python
Java

Messaging & composition

NATS
Docker Compose

Evidence

138 automated tests.
Clean linux/amd64 builds.
Bounded six-service smoke testing and verified cleanup.

Limitation

The profitability gate was not proven to pass in the challenge.

View case study →

A Python diagnostic exercise covering domain, DNS-zone, and record retrieval, private A-record diagnosis, and hostname-to-public-IP verification.

Stack

Implementation

Python standard library

Interfaces

HTTP APIs
DNS

Evidence

Retrieved domain, DNS-zone, and record data through HTTP APIs.
Diagnosed a private A record and checked hostname-to-public-IP resolution.
Produced a specific DNS correction recommendation.

Limitation

This was a technical hiring challenge, not employment.

View case study →

Additional work

Python, TypeScript and C# contributions spanning MCP approvals, permission-state consistency, cancellation cleanup and wallpaper recovery.

Stack

Python
TypeScript
C#
MCP
asyncio
SQLite

Evidence

Eight authored PRs merged as of 24 October 2026, including persistent approval policy and exact-call Pi MCP grants (#234), and wallpaper rollback/restart recovery (#228).
Cancellation fix #226 added three regressions that failed before the repair and passed afterward within a related 28-test suite.

Limitation

Extension update/rollback PR #261 remains open. Contribution does not imply project ownership or a maintainer appointment.

Public evidenceAuthored Amadeus pull requests Public evidencePending extension-update proposal #261

A C#/.NET 8 TCP game server and JavaScript browser chess client with shared state, concurrency controls, validation, and logging.

Stack

C#
.NET 8
TCP sockets
JavaScript
REST-style requests

Evidence

Implemented networked server/client concepts with shared state and validation.

Limitation

Presented as a compact supporting project rather than a featured case study.

A Python analysis of water-quality data covering cleaning, exploration, model comparison, visualisation, and evaluation.

Stack

Python
pandas
scikit-learn
SPSS

Evidence

Used cleaning, exploratory analysis, model comparison, visualisation, and evaluation.

Limitation

A focused analysis project rather than a deployed prediction service.

A stakeholder-oriented workflow analysis with SMART KPIs, cost-benefit analysis, and an improvement proposal.

Stack

Workflow analysis
SMART KPIs
Cost-benefit analysis

Evidence

Produced stakeholder-oriented improvement recommendations.

Limitation

An analysis and proposal, not evidence of organisational implementation.

A clean-room Java domain core for event-driven trading systems, covering execution-based accounting, inventory-aware quoting, and aggregate risk controls.

Stack

Java 17
Maven
JUnit
GitHub Actions

Evidence

46 focused JUnit tests: 7 accounting/model, 19 quoting/lifecycle, and 20 aggregate-risk checks.
Three transport-neutral modules: common, quoter, and hedger; no exchange connection is required.
Validation documents distinguish the public core from historical challenge-system tests and simulations.

Limitation

The public core does not connect to an exchange or submit orders. Historical challenge results are separate; profitability and production trading performance are not established.

Public evidencePublic source and reproducibility guide

04 / EXPERIENCE & EDUCATION

Learning through delivery

Amadeus · Open source
October 2026 — Present
Open Source Contributor
Eight merged PRs in Python, TypeScript and C# covering MCP approval, permission state, cancellation cleanup and wallpaper recovery.
Implemented persistent approval policy and exact-call grants while preserving permission boundaries; extension update/rollback proposal #261 is pending merge.
University of Auckland Industry Internship
October 2025 — March 2026
Frontend & AI Developer Intern
Built reusable Next.js and React components to improve consistency and reusability across project pages.
Integrated frontend workflows with REST APIs, clarified contracts with backend developers, and supported stable service data flow.
Developed prompt templates and search logic; contributed to sprint planning, code review, testing, documentation, and team delivery.
University of Auckland
March 2025 — September 2026
Master of Information Technology
First Class Honours · GPA 7.417 / 9.0 · Final-semester GPA 8.0
Coursework: Data Analytics with Python/SPSS · Big Data Algorithms · IT Consulting with SAP tools
Nanjing Forestry University
September 2020 — June 2024
BSc Information Management and Information Systems
GPA 3.4 / 4.5 · Top 20%

05 / PROFILE

How I work

Observe before acting
Start from current evidence rather than assumptions.
Bound the decision
Make ownership, risk, and allowed actions explicit.
Verify the outcome
Distinguish attempted work from proven results with tests, traces, reconciliation, and stated limitations.

Technical skills

Languages

Python
TypeScript
JavaScript
C#
Java
SQL
HTML/CSS

Frontend

React
Next.js
Tailwind CSS
Responsive UI
Fetch API

Backend & APIs

FastAPI
ASP.NET Core Minimal API
REST APIs
JWT
API integration

AI & Automation

LLM workflows
MCP
OCR
Vision-language models
Structured outputs
Prompt engineering

Data & Databases

MySQL
SQLite
MongoDB
Neo4j
pandas
scikit-learn
SPSS

Testing & Delivery

pytest
JUnit
Jest
Playwright
Git
GitHub Actions
Docker
CI/CD

Systems

HTTP
TCP sockets
Multithreading
Concurrency control

Hi, I'm Wenqing.I build reliable software for uncertain systems.

Stack

Evidence

Limitation

Stack

Evidence

Limitation

Stack

Evidence

Limitation

Stack

Evidence

Limitation

Additional work

Stack

Evidence

Limitation

Stack

Evidence

Limitation

Stack

Evidence

Limitation

Stack

Evidence

Limitation

Stack

Evidence

Limitation

Open Source Contributor

Frontend & AI Developer Intern

Master of Information Technology

BSc Information Management and Information Systems

Observe before acting

Bound the decision

Verify the outcome

Technical skills

Languages

Frontend

Backend & APIs

AI & Automation

Data & Databases

Testing & Delivery

Systems