Skip to content
中EN
Home01Education02Experience03Projects04Contact05

Hi, I’m

Luke Yao姚宇柯

Passionate about AI products with a strong drive to build ideas into hands-on prototypes.

See my resumeGet in touch
BOSS ZhipinLinkedInEmail
Luke Yao

Ask my AI twin

Answers from my resume and project docs

Beta
01What I do

What I do

Fig. 01
01 / 03

AI Product

Starting from real business scenarios, I work out where and how AI should fit in, design the solution, and evaluate it to prove it works.

AxureDifyLangGraphRAGMCPHarness Engineering
  • 01Business understandingHelped design a company OA system from scratch, and can spot where AI fits in a business process and write the PRD for it.
  • 02Solution designBuild multi-step workflows in Dify or LangGraph, deciding what the model handles and what stays with hard rules.
  • 03Harness engineeringFamiliar with agent harness design; built ReAct agents and package AI capabilities as tools via MCP.
  • 04Evaluation-drivenBuild eval sets and let quantitative results guide each round of optimisation.
  • 05SafeguardsDesign validation checks and human review to catch hallucinations and errors.
Fig. 02
02 / 03

Software Engineering

With a computer science degree and hands-on front-end and back-end experience, I can read a technical design, judge what it will cost to build, and put together a working prototype myself.

JavaPythonTypeScriptVueMySQLDocker
  • 01Engineering fluencyFront-end and back-end internships taught me how a feature travels from the UI to the database, so I can speak the same language as engineers in reviews.
  • 02Cross-domain deliveryBuilt experiment software on my own for a neuroscience research team, turning the paper's mathematical paradigms into reproducible programs; the work supported a published SCI paper.
  • 03Rapid prototypingI use AI coding tools like Claude Code to build working prototypes quickly — this website is one of them.
Fig. 03
03 / 03

Robotics

MSc in Robotics, spanning perception, control and human–robot interaction.

ArduinoMATLABSimulinkYOLOOpenCV
  • 01Human–robot interactionDesign user studies that measure how much people trust and accept intelligent systems.
  • 02Computer visionCombine classic image processing with deep learning models, covering the full pipeline from data processing and model training to evaluation.
  • 03Simulation & validationModel and simulate control systems, and design controlled experiments to test how robust a solution is across conditions.