Heng Xu

Master of Computer Science candidate at the University of Sydney with a BEng background in aircraft manufacturing engineering, focused on application delivery, API/UAT support, release readiness, and AI-assisted implementation.

Experience includes FinX, Luna, and a team USYD Capstone PII redaction API, with individual contributions covering service workflows, API validation, benchmark reporting, deployment checks, and handover documentation.

Application Delivery and API/UAT Support Intern - Finfo Global - on-site then remote

(2025-12)

WeChat Mini Program + Django/DRF API | project-based

• Delivered FinX Mini Program implementation and QA/UAT handover, converting registration, ticketing, payment preparation, check-in, refund/transfer, content, and IM workflows into API checks and acceptance criteria.
• Used Cursor/Codex/Claude while retaining review ownership; validated idempotency, payment callbacks, ticket/refund states, verifier permissions, error handling, release gates, and handover notes.
• Supported Tencent Cloud CVM readiness: Nginx/Gunicorn/systemd checks, log paths, verification scripts, WeChat Pay preparation, and release-gate documentation.

FastAPI Service Developer - USYD Capstone

(2026-03)

AI Hub PII Redaction API Prototype | API contract | OCR/file ingestion | benchmark reporting | client-facing documentation

• Individually owned the data pipeline, backend API, testing/benchmark, deployment, and operations workstreams in a team capstone, turning the PII workflow into a service pipeline for text/file input, parsing/OCR, model detection, span post-processing, block/review policy, and redacted output.
• Documented FastAPI/OpenAPI contracts, OCR/file metadata, pytest/CI checks, and AI TOP/Tailscale deployment validation; consolidated benchmark results across F1, parse rate, cost, latency, and throughput.

AI Integration Developer - Luna

(2025-11 - 2025-12)

Local Desktop AI Assistant Prototype v0.6 | FastAPI + Electron | local AI tooling | runtime validation

• Built the AI-integration layer with AI-agent assistance, connecting Electron and FastAPI with llama.cpp-served Qwen2 GGUF models, XTTS v2 services, and local/cloud LLM routing.
• Packaged the v0.6 runtime for repeatable local use, including portable data isolation, startup orchestration, health checks, and runtime/config validation.

Technical Intern - China Eastern Technology Application R&D Center

(2023-06 - 2023-07)

• Built exposure to regulated maintenance workflows through A320/B737 structure and systems training, ERe support, and MD-11 materials record updates.

Research Assistant - Research

(2024-11 - 2025-01)

Kolmogorov-Arnold Network Application

• Used KAN to model and compare signal functions with PySR, focusing on interpretability in time-domain analysis and symbolic regression.

Bachelor Thesis Author - Shanghai University of Engineering Science

(2023-12 - 2024-05)

Gear Fault Diagnosis with RPM-driven Deep Learning

• Converted vibration-signal data into RPM-based image representations in MATLAB and supported testing of a deep-learning fault diagnosis model, with thesis experiments reporting 90%+ diagnostic accuracy.

Master of Computer Science - Computer Science - University of Sydney (2024-07 - 2026-07)

BEng - Aircraft Manufacturing Engineering - Shanghai University of Engineering Science (2020-09 - 2024-06)