Get in touch: hello@rpascale.org · LinkedIn
Hui Fang spent over a decade in Singapore's defence research sector as an algorithms and systems engineer, working on Computer Vision and Machine Learning. Not the research-paper kind of ML — the kind where the pipeline has to actually run, the detections have to hold up on real sensor data, and someone is waiting on the other end for a system that works. She went through enough full development cycles to develop a healthy scepticism of anything that only works in the notebook.
The work covered the usual range: object detection and tracking in images and video, ML infrastructure for training and evaluation, system integration across components, and eventually leading projects end-to-end from design through to operational deployment. One of her works on an Advanced Imagery Management System was recognised with the Defence Technology Prize in 2023.
In 2024 she started an independent practice helping small businesses with AI and automation. The context is different from defence but the core problem is the same: figure out what a process actually needs, then build something that doesn't fall apart when real people use it. She's worked on agentic systems for business messaging and CRM, business process automation, LLM evaluation frameworks, and quantitative research tooling. She is particular about sound software engineering fundamentals even when her LLMs are doing most of the coding. She still goes in and tinkers directly when dissatisfied, and gives her agents a good scolding afterwards, hoping it'll imprint into their memory somehow.
She thinks about these problems from the technical and business side at the same time, mostly because she's found that purely technical solutions often miss what the business actually cares about, and purely business-framed ones tend to underestimate what it takes to build something robust. The sweet spot is usually somewhere in between, which is also where the interesting work tends to be.
She's probably a reasonably good company if you want to talk about tech, investments, AI agents, or whatever half-baked idea you've been turning over. She has opinions, but she'll hear yours out first.
This page was largely written by an LLM that knows her pretty well. She approved it, mostly.