Featured work
Live projects & published research
↗
Live demo
Weather Analytics Dashboard
Real-time Superset dashboard · 10 cities · 30-min refresh
Airflow · dbt · PostgreSQL · Azure VM · Cloudflare Tunnel
↗
Published research
Mental Health Topic Modelling — ICRTCIS 2025
First-author · Springer Nature Proceedings 2026
MentalBERT · BERTopic · LangChain RAG · 10,000+ posts
Data Engineer & ML Practitioner

Rain
Liao

Master of Data Science graduate from the University of Adelaide. Passionate about building scalable data systems, cloud infrastructure, and turning raw data into actionable insight.

View Projects →
3
MS Certifications
5+
Projects Shipped
Rain Liao
Rain Liao — casual
off the clock ★
PythonSQLAirflowdbtDocker AWSAzurePySparkNLPPyTorch LangChainPower BIQGISDynamoDBFreqtrade PythonSQLAirflowdbtDocker AWSAzurePySparkNLPPyTorch LangChainPower BIQGISDynamoDBFreqtrade
Who I am

Curious by nature,
rigorous by practice

Data Engineer with hands-on experience designing and deploying scalable ETL pipelines, cloud infrastructure, and data storage solutions across AWS and Azure. I enjoy building systems that make data usable — from ingestion and transformation through to structured storage and reliable delivery.

I'm comfortable writing production Python, wrangling SQL, and spinning up containerised environments that teams can depend on. My background in civil engineering gave me a strong analytical foundation that carries directly into how I approach data problems today.

Outside of work, I enjoy jogging, weight training, and swimming to stay active, and I've recently been getting into wine and bartending — which keeps things interesting away from the screen.

Data Engineering NLP & LLMs Cloud Infrastructure Quant Finance Full-Stack
Languages
Python · SQL · R · C++ · JavaScript
Data Engineering
ETL/ELT · Airflow · dbt · PySpark · Batch & Realtime · Data Quality
Cloud & DevOps
AWS (EC2, Lambda, S3, DynamoDB, CloudWatch) · Azure VM · Blob · Docker · Linux
ML / AI
PyTorch · Scikit-learn · LangChain RAG · Transformers · BERTopic
Data & Viz
Power BI · QGIS · Superset · Pandas · Selenium
Web
React · Vite · Django REST · REST API Design
Career

Experience

2025
Univ. of Adelaide
Data Science Intern — SET Faculty
Designed and deployed an end-to-end ETL pipeline (Python, Selenium, NLTK, Azure Blob, Airflow) to ingest, clean, and structure 10,000+ unstructured social media posts into a query-ready format — applying statistical validation to topic modelling that surfaced 10 key mental health subtopics, and engineering a scalable processing workflow supporting transformer-based models and a LangChain RAG system for downstream NLP analysis. Collaborated with academic supervisors and contributed as first author of a conference paper accepted at ICRTCIS 2025.
PythonAirflowAzure BlobSeleniumNLTKLangChainRAG
2024
AIML
Quantitative Trading Assistant — AIML
Built and maintained a high-volume data pipeline processing 100GB+ of Binance market data, combining fine-tuned LLM sentiment signals with technical indicators for automated trading strategy execution using Python and Freqtrade — then containerised the full research environment with Docker and deployed it to Azure VM, cutting environment setup time by 80% and delivering a reproducible, production-ready pipeline for live trade execution.
PythonFreqtradeDockerAzure VMLLM Sentiment
2023
NCREE Taiwan
Special Case Technician — NCREE
Scraped, processed, and structured large-scale weather and seismic datasets using Pandas and QGIS, storing results reliably in AWS DynamoDB and S3 for scalable downstream access — then built and deployed a real-time data ingestion and alerting system on AWS (EC2, Lambda) integrating REST APIs and Google Maps API, with end-to-end observability via CloudWatch and SNS on a Linux-based infrastructure, coordinating a cross-functional intern team to deliver the system on schedule.
AWS EC2LambdaDynamoDBS3CloudWatchQGISSNS
2025 –
Olive Adelaide
Front of House — Olive Fine Dining
Delivering excellent guest service through clear communication, teamwork, and quick problem-solving in a fast-paced fine dining environment — sharpening interpersonal and client-facing skills alongside a technical career.
CommunicationTeamworkClient Service
Credentials

Education &
Achievements

Feb 2024 – Dec 2025
Master of Data Science
The University of Adelaide, Australia
Sep 2018 – Jun 2023
Bachelor of Science in Civil Engineering
Chung Yuan Christian University, Taiwan
Languages
English  ·  IELTS 7.5
Mandarin — Native
DP-700
Fabric Data Engineer Associate
Microsoft · DP-700
DP-100
Azure Data Scientist Associate
Microsoft · DP-100
PL-300
Power BI Data Analyst Associate
Microsoft · PL-300 · 2025
📄
First-Author Conference Publication
ICRTCIS 2025 · Springer Nature · Proceedings 2026
🏆
ALTA 2025 — ADR Detection
Ranked 5th / 150 participants · +20% accuracy over baseline
Portfolio

Selected Projects

01
Personal · 2026 · Live
Real-Time Weather Analytics Pipeline
Airflow 3.0 · dbt 1.9 · PostgreSQL 14 · Superset 3.0 · Docker Compose · GitHub Actions · Azure VM · Cloudflare Tunnel
Weather API → Airflow → PostgreSQL → dbt → Superset → Cloudflare → Public
10
Cities
30m
Refresh
6
Services
3
dbt Layers
Production weather ETL pipeline — six containerised services (PostgreSQL, Airflow, dbt, Superset, Redis, Cloudflare Tunnel) orchestrated by Docker Compose on Azure VM. Data moves through a medallion architecture: staging → silver (deduplication, UTC-to-local TZ conversion) → mart (daily averages, hourly detail, city summaries). A GitHub Actions CI/CD pipeline validates DAGs, runs dbt parse checks, and executes pytest on every branch before deploying via SSH — credentials injected from GitHub Secrets at runtime, never committed to the repo. Published publicly via Cloudflare Tunnel: outbound-only connection means zero open firewall ports, no certificate management, and built-in DDoS protection.
Live · superset.rainsunny.org
Weather Analytics Dashboard Preview
Open Live Dashboard
GitHub → Live Demo ↗
↗
02
Published · Springer Nature
Mental Health Topic Modelling
Python · MentalBERT · BERTopic · Network Analysis · LangChain
Subtopic network graph — mental health forum themes
Applied MentalBERT-enhanced BERTopic to 10,000+ Beyond Blue forum posts, surfacing 10 recurring themes around medication, therapy, and work stress. Built a visual subtopic network revealing how Australians express distress and engage in peer support — with a LangChain RAG chatbot for context-aware responses. Accepted at ICRTCIS 2025, published in Springer Nature proceedings 2026.
GitHub → Publication ↗
↗
03
Algo Trading · In Progress
Binance Delisting Strategy
Python · python-binance · Selenium · BeautifulSoup · pandas
Automated pipeline that detects Binance delisting announcements and opens risk-managed short positions on the affected pairs. A Selenium-based crawler scrapes the delisting board and diffs new announcements against previously-seen ones; a margin trading module sizes positions from available balance and opens isolated-margin shorts via the Binance API. Entry logic and position sizing are implemented and tested against historical announcement data; exit-side stop-loss/take-profit monitoring is still being refined.
View Code → ↗
04
NLP · Coursework
NLP Classification Portfolio
Python · scikit-learn · NLTK · spaCy · Sentence-Transformers · BERTopic
Two NLP projects built during the Master of Data Science. A hotel review sentiment classifier compares TF-IDF/Naive Bayes against the VADER lexicon and a hybrid model, extending into aspect-level sentiment with a multi-output classifier. A Stack Overflow post categoriser collects questions via the Stack Exchange API and groups them into 12 NLP task categories, comparing a keyword-based approach against SBERT/CodeBERT embeddings with BERTopic-driven topic discovery.
View Code → ↗