Relia ↗
A hosted API that scores how relevant a news article actually is to a company, not just whether it's mentioned.
- Relia-1 · LoRA Qwen3-Embedding bi-encoder
- 86.5% agreement with frontier LLM raters
- ~18ms / article
Design engineer and founder, based in London
I build the models, the pipeline, and the product behind Aleta Index, the data infrastructure that turns the world's news into credible, structured signal for AI agents and decision-makers.
About
Xiaoqin “Jamie” Zhao. I trained across the Royal College of Art (MA, Innovation Design Engineering) and Imperial College London (MSc, Design Engineering, with a Machine Learning elective), after a Bachelor of Business.
That mix is the point: I sit where ML, product, and design meet. At Aleta Index I lead everything from LoRA fine-tuning and data pipelines to information architecture and the product people actually touch.
I think best in motion. Most of my hardest problems get solved on a long bike ride, not at a desk.
What I'm building
A narrative intelligence company. We track how claims spread across the world's news and score which sources are actually credible, turning noise into structured signal you can trust.
A hosted API that scores how relevant a news article actually is to a company, not just whether it's mentioned.
Gives an AI agent a financial narrative brain: clusters news into narratives, tracks sentiment momentum, maps narrative risk.
The live news-narrative-intelligence product, powered by a Claim DAG and an RL-based source credibility engine. Owned end to end, architecture to launch.
Experience
Aleta Index, London
Led product architecture, prototyping and full-stack build of the Narrative Intelligence Platform (B2B API and MCP server). Built the end-to-end data pipeline: LoRA embedding fine-tuning on financial corpora (PyTorch Lightning, GCP), narrative modelling, and a source-credibility engine (Claim DAG with RL scoring). Direct the technical team across every workstream.
Hercules Cambridge Design Studio, London
Directed research-driven design with multidisciplinary student teams. Led end-to-end behavioural research for Sightsavers (UK NGO) on how younger donors evaluate trust and impact in digital giving.
ARPA-Cell Protocol (listed on Coinbase & Binance), remote
Designed grid-trading system architecture and market-maker mechanics. Analysed on-chain user behaviour to cut friction in transaction flows and shorten decision paths.
JD.com (Jingdong), Beijing
Modelled GMV and retention dynamics, turning transactional data into category-level strategy and decision-oriented insight for non-technical stakeholders.
Education
MSc, Design Engineering, with a Machine Learning elective
MA, Innovation Design Engineering
Bachelor of Business
Recognition
Toolkit
Contact
Building something at the intersection of markets, AI and design, or just want to compare notes? I'd love to hear from you.