Clay Liang

Clay Liang 梁珂玮

claysps@outlook.com

I work at the intersection of data science and decision-making — building tools, dashboards, and models that bridge the gap between raw data and the people who act on it.

Currently a BSc student at Leiden University (AI & Data Science), heading to TU Munich next. Previously interned at Airbus (data analytics, Tianjin) and Lenovo (data science), and doing research at Tsinghua University on deep learning for audio.

CV / GitHub / LinkedIn / Email

Blog

Data, markets & thinking

Projects

Things I've built

Fin Tools

Market data & analysis

Reading

Books & annual reports

Experience

Data Analyst Intern, Airbus (Tianjin Final Assembly)
Jun 2025 – Sep 2025

Used SQL and Apps Script to query operational data, surface decarbonisation metrics and process bottlenecks
Built Databricks / Python workflows for large-scale preprocessing
Main owner of visual outputs — charts, infographics, slide decks — delivered to cross-functional teams

Data Science Intern, Lenovo
Feb 2025 – Jun 2025

Designed Tableau & Power BI dashboards for digital transformation strategy
Developed training data optimisation strategies to improve model performance
Built Excel / PowerPoint templates for technical presentations; led intern team as project lead

Research Assistant, Tsinghua University
Jun 2024 – Present

Co-developed deep learning models for audio processing; improved speech recognition accuracy via data augmentation and loss function tuning
Managed and structured datasets for training and evaluation
Debugged backend code to reduce latency and improve runtime efficiency

Projects

🎥

Video Annotation Tool

Local video annotation tool with browser-based UI. Supports playback, timestamped labeling, and CSV export for ML datasets. Built for small-scale research workflows where off-the-shelf tools are overkill.

Python · Flask · OpenCV · JavaScript

🎵

Offline Music Recommendation

Emotion-aware music recommendation system with speech recognition input and local playback. Combines sentiment analysis with a content-based recommender — entirely offline, no API dependencies.

Python · Speech Recognition · Emotion Analysis

⚡

csvtool

Lightweight CSV parser in C++. Written because Python was too slow for the dataset size at the time. Focused on performance and memory efficiency.

C++

🌍

AI-Eco

Interactive web prototype visualising the environmental impact of AI models — energy consumption, water usage, e-waste. Built with Three.js and Chart.js. Course project on interaction design.

HTML · JavaScript · Three.js · Chart.js

I also built a set of free financial tools — live FX, crypto, stock quotes, DCF valuation, and an ASML tracker that plots the market price against my own DCF estimate. Details on the projects page.

Education

BSc Management & Data Science · TU Munich · 2026 –

BSc Artificial Intelligence & Data Science · Leiden University · 2025 – 2026

Skills

Languages: Python, SQL, C++, Java, JavaScript

ML / DL: PyTorch, TensorFlow, Scikit-learn, Pandas, NumPy

Data & Infra: Databricks, Docker, Git, Linux, Spark

Visualisation: Tableau, Power BI, Matplotlib, Seaborn

Audio: Librosa, FFmpeg, Praat

Office & Comms: Excel (advanced), PowerPoint, Google Workspace

Miscellaneous

I read annual reports recreationally, which is either a personality trait or a warning sign. Also: photography, Chinese cooking, and enough German to survive a Bäckerei. Currently reading Damodaran and wondering why I didn't start earlier.

Looking for internships where data touches money — banking, asset management, fintech, risk. If that's your world and you're open to a conversation, I'd like to hear from you.