Xi Zhang
Founder & AI Engineer · AI Business Automation
Melbourne, AU
I'm an AI engineer and aspiring founder. My focus is agent-based business automation for Australian SMEs — turning the manual back-office workflows that previously required full-time staff into autonomous AI workflows that customers can adopt in days.
I have already taken AI systems from ambiguous requirements through architecture, training, optimisation, and production hand-off twice over for Fortune 500 clients — not research demos, but AI-based applications that customers paid for and integrated into their operations.
Alongside that work, I'm a Postdoctoral Research Fellow at Macquarie University. My PhD (RMIT, 2024) produced research published at MobiCom, SenSys, IEEE TMC, IEEE/ACM ToN and other CORE A* venues — depth that now informs the platform I am building for SMEs.
Experience
Founder & AI Engineer
AI Business Automation Practice · Melbourne, AU
- Building an autonomous agent orchestration stack on top of OpenClaw, Hermes, OpenGOAL, and MCP (Model Context Protocol) — the technical foundation for AI-augmented business workflows
- Developing reusable patterns for multi-agent coordination, tool use, and DAG-style workflow execution, productised for enterprise customers
- Engaging Australian SME clients to design and deploy AI-driven workflow automation — replacing labour-intensive back-office processes with autonomous agents
Postdoctoral Research Fellow
Macquarie University · Sydney, AU
- Research on multimodal AI and on-device inference for real-world interactive systems
- First-author of LargeCall (IEEE INFOCOM 2026, CORE A*) integrating LLMs with smartphone sensor data for real-time speech enhancement
- Guest lecturer for COMP8230, COMP8296, and COMP3210/6210; supporting HDR student supervision in Data Science and AI
AI Engineer · Tech Lead
Deego Technology · Melbourne, AU
- Designed and built a full AI platform for a Fortune 500 retail client — phone-sensor-only indoor positioning across >10,000 m² stores, fully automated end-to-end with no cameras and no human-in-the-loop
- Built autonomous AI agents for each pipeline stage (sensor fusion, posture classification, trajectory prediction), composed into an end-to-end workflow rather than a monolithic model
- Owned the full ML lifecycle: architecture, training, validation against drift targets (<3 m over 20 min), on-device optimisation (quantisation, pruning), Docker + CI/CD
- Delivered a production AI-based application that the client's engineering team integrated directly. Set technical direction as tech lead
AI Engineer (Research & Design)
RMIT University · Melbourne, AU
- Led greenfield AI development for a Fortune 500 client's smartwatch gesture recognition system — replaced a failing classical pipeline with deep learning, lifting accuracy to 95%+ across 8 gesture classes, 6 scenarios, 20 participants
- Reduced inference latency by 50%+ through model compression and system profiling, enabling real-time deployment on resource-constrained hardware
- Built a fully reproducible end-to-end workflow (data collection, preprocessing, training, hyperparameter tuning, validation) ready for handoff
Education
PhD · Computer Science
RMIT University · Melbourne, AU
Thesis: Enabling Advanced Human Sensing through Millimetre-wave Radar with Deep Learning
9 publications across CORE A* venues
Master of Information Technology · Computer Science
Monash University · Melbourne, AU
Thesis: Anomaly Detection in Spark-Scala for Big Data
Information Technology International Merit Scholarship
Bachelor of Computer Science · Computer Science
Beijing Jiaotong University Haibin College · Beijing, CN