AI Operations · Agent Workflows · Finance Systems

MartinZhang

Looking for
AI Ops · AI Product · Agent workflows
Available nowToronto · RemoteNow buildingJazzTor · War Trader
Stylized 3D portrait of Martin Zhang smiling and wavingSay hi 👋
15K+Automated tests passing across my games
31K+Lines of code shipped in JazzTor and War Trader
550K+Views on my AI-generated video work
173Music tracks and sound effects synthesized in code
200+Mortgage payout statements validated monthly
  • AI operations
  • Agent workflows
  • Human-in-the-loop review
  • Test-driven delivery
  • Finance controls
  • Godot games
  • Real-time 3D
  • Bilingual EN / 中文

Selected work

Seven projects. One way of working.

All projects

01 / Experience

Professional experience

Apr 2026 — Present

Payout Administrator

Equitable Bank Toronto, ON

  • Highlight 01200+ payout statements a monthprepared and checked for residential and commercial mortgages, on tight closing deadlines.
  • Highlight 02Exception handlinginvestigate variances in balances, per-diem interest, penalties, and fees before a statement goes out.
  • Highlight 03100+ partnerslawyers, brokers, and internal teams I coordinate with to resolve discrepancies and close on time.

Mar 2025 — Apr 2026

Legal & Executive Operations Coordinator

JYD LLP Toronto, ON

  • Highlight 01Right hand to two founding partnersacross the firm’s legal and mortgage practices.
  • Highlight 02100+ active matterskept on track through deadlines, transaction packages, and client correspondence.
  • Highlight 031,000+ recordsmaintained in Clio and Salesforce for compliance, trust accounting, and billing.

Jun 2023 — Aug 2023

Client Manager Intern — Commercial Real Estate

Jianghai Securities Co., Ltd. Shanghai, China

  • Highlight 0150+ clients · 20+ financing casessupported with financial review, market research, and case tracking.
  • Highlight 02300+ client touchpoints a weekin Mandarin and English, with clean CRM records and follow-ups.

Ongoing · alongside full-time work

Independent AI Practice

Self-directed Toronto / Remote

  • Highlight 0110K+ likes · 550K+ viewson AI video work I scripted, generated, edited, and published on RedNote.
  • Highlight 02Hands-on model testingcomparing video and image models on prompt following, motion, continuity, and character consistency.
  • Highlight 03AI product testingturning real scripts into test paths, logging agent failures, and writing prioritized fixes.
Beyond the deskPart-time & volunteer experienceHospitality, events, and community work: 3 roles since 2022.

02 / Skills

What I work with

AI systems & agents

Designing where AI helps, where rules decide, and where a person signs off.

  • Claude Code Skills & agent teams
  • Agent instructions & structured outputs
  • Human-in-the-loop review
  • Evaluation (DeepEval)
  • Multi-model testing
  • RAG fundamentals
  • Local models (Ollama)
  • Python · SQL
8 tools

Product & build

Taking a fuzzy idea to a prototype people can click.

  • Problem framing & PRDs
  • User flows & edge states
  • React · TypeScript
  • Figma
  • Accessibility review
  • Git · Docker
6 tools

Games & 3D

Where I stress-test AI-directed production end to end.

  • Godot · GDScript
  • React Three Fiber · three.js
  • Blender Python pipelines
  • Procedural audio
  • Test-driven game QA
  • Playwright visual QA
6 tools

Finance operations

The domain my AI work is grounded in.

  • Reconciliation & exceptions
  • Month-end close controls
  • Journal entry preparation
  • Driver-based forecasts & variance analysis
  • Internal controls & audit trail
  • Structured data contracts & validation
  • Excel · Power BI · SQL · Python
  • English · Mandarin
8 tools

03 / Built Skills

AI Skills I’ve built

Reusable agent workflows with clear inputs and outputs someone else can check.

Built · Iterating

GEO Brand Audit

Measures how often AI assistants mention and cite a brand, before anyone suggests changes.

Input
Search-intent queries · AI answers · brand mentions · cited sources
Output
Visibility matrix · query-level log · citation gaps
View source
Built · Iterating

GEO Brand Optimizer

Turns the audit’s gaps into a prioritized fix list, then re-runs the same test.

Input
Audit findings · crawl settings · page content gaps
Output
Action backlog · implementation brief · before/after delta
View source
Built · Iterating

Product Design Mentor

An AI tutor that teaches product design through practice, critique, and revision.

Input
Learner goal · design draft · user context · constraints
Output
Focused lesson · critique · exercise · revision plan
View source
Built · Iterating

Finance Exception Review Agent

Explains why an Accounting or payroll control failed—without calculating, posting, or approving money.

Input
Rule code · cited evidence · severity · allowed actions
Output
Grounded brief · next action · open questions · model telemetry
View source

04 / Education

Education

Academic recordGraduated · Dec 2024

Honours Bachelor of Arts

University of Toronto Mississauga
EconomicsCommunication, Culture, Information & Technology
  1. Honours Bachelor of ArtsEconomics · Communication, Culture, Information & Technology
  2. CFA Level I candidateFinance
  3. CSC self-studyMarkets

Certifications

Licences and certificates, newest first.
  • Smart Serve OntarioResponsible alcohol serviceCertified

Study threads

Interdisciplinary by design
  • Economics
  • Commerce
  • Digital finance
  • Financial analysis
  • Data analysis
  • Web design
  • UI design
  • Project management
  • Process design

05 / Work

See how I think when I build.

AI controls, responsible-AI tools, a 3D jazz room, a mobile app, and two games.

Portfolio / WorkEnter the archive