Applied AI Engineer · Agent Systems, AI Infrastructure & DBMS
Hi, I’m Jason Wang, a Sydney-based applied AI engineer focused on agent systems and infrastructure. I also tutor at the University of Sydney’s School of Computer Science.
I’m the lead maintainer of NoKV, an open-source infrastructure project providing persistent workspaces and state management for AI agents. NoKV is part of the Linux Foundation ecosystem, listed in both the LF AI & Data and CNCF landscapes, and is also catalogued in CMU’s Database of Databases (DBDB.io) by Prof. Andy Pavlo. NoKV is actively collaborating with LoopX (5.8K+ GitHub stars), OpenViking (37K+ stars), and Hermes Agent (245K+ stars).
I’ve been building with generative AI since late 2023 and have helped two local startups develop customer-facing AI agent systems from scratch. At THDR Group, I implemented the core agent workflow for Neuono, an AI fashion platform later featured in Forbes Australia and showcased at New York Fashion Week in September 2025.
My capstone project, EchoJournal, is a voice-native personal memory system powered by a real-time voice model. It was selected for the University of Sydney’s Genesis Accelerator, Cohort 36, in late 2025.
I value the combination of product development and infrastructure engineering in practical AI applications, particularly in industries beyond tech. I’m especially interested in working closely with domain experts to understand their workflows and turn their needs into useful, reliable agent systems.
Fun fact: I was a journalist back in 2019.
Co-founder
Co-built an agent-native filesystem and versioned state layer so workspaces can outlive their sandboxes. My work spans engineering, downstream use-case discovery and open-source collaboration.
Listed in LF AI & Data LandscapeCNCF Landscape
Founding Engineer · Contract
Built the data and retrieval architecture for a fashion agent harness, including the production-database-to-Graph-RAG ETL pipeline and graph design. Used authoritative garment metadata and images to constrain personalised product previews and connect design to ordering.
Showcased at New York Fashion Week 25SS.
Full-stack Engineer · Contract
Independently shipped v1.0 of a photographer-focused image management and sharing product. Owned the application flow from uploads and storage through shareable galleries and billing.
100+ daily active users and 30+ paying users in the first month.
Collaborator
Contributed shared-goal state and execution-ownership semantics to a long-task agent harness. The work makes authority explicit: who may act and publish results, and how state stays consistent across retries, reassignment and recovery.
5.8K GitHub stars (as of 2026-09-13).
Personal project
Built a voice-native personal memory system that turns spoken entries into a searchable journal and structured reflections. Added real-time voice conversations grounded in the user’s own history, with explicit time-range retrieval.
University of Sydney Genesis Accelerator, cohort 36.
