Yixin Tian [田毅鑫]
Computer Engineer & Data Scientist

Hey! I'm Yixin ("Isshin"), a computer engineer and data scientist based in Toronto, Canada. I design and build human-centred AI/ML systems at the intersection of software engineering and ML research.
Currently, I work as a Data Scientist at Royal Bank of Canada (RBC), where I focus on developing agentic AI assistants for the credit domain, leveraging DSPy, LangGraph, and tool-augmented LLMs. In the summer of 2025, I served as a team lead for RBC Amplify, mentoring a team of four to rapidly prototype an AI assistant for account managers using retrieval-augmented generation (RAG), text-to-SQL LLM agents, and automated client insight generation, powered by anomaly detection. My previous experience spans deploying the open-source Dagster data orchestration platform, building event-driven inference and monitoring pipelines, and co-developing a patent-pending agriculture carbon-emission model into production.
Outside of work, I'm a part-time researcher at the University of Southern California's Information Sciences Institute (ISI), where I study structured data extraction (text-to-JSON) from long documents, knowledge graphs, and LLM agents.
My passion lies in exploring better modes of knowledge sharing to expand the bandwidth of human–AI, human–human, and AI–AI communication in real-world cooperation and education. When I'm not building or researching, I enjoy running, writing, reading, and hiking in nature.
Oct 4, 2025
The hitchhiker's guide to AI coding tools
Apr 16, 2025
Multi-agent debate with state pattern from scratch
Mar 4, 2025
Book Review - Nexus by Yuval Harari
Dec 28, 2024
Introduction to Ontology
Jul 7, 2024
Reflect on the three dimensions of effective learning
Jun 30, 2024
Migrate personal blog to Next.js + MDX
Jan 14, 2024
When evolution meets art (text-to-image via CLIP)
Nov 11, 2023
The problems of modern note-taking apps
Jul 14, 2023
Understand linear regression through many facets
Jun 4, 2023
Question answering over multiple documents using LLM
May 3, 2023
A first look at learning, from a not-so-fast learner