Hi there!👋🏼

I’m Tri Watanasuparp, a data scientist specializing in machine learning, NLP, and building intelligent systems that turn complex data into actionable insights.

Currently pursuing an M.S. in Data Science at Northeastern University

watanasuparp.t@northeastern.edu

experience

Lightcast

Sep 2025 – Present

Data Analyst (Applied NLP)

• Built and optimized NLP pipelines to classify Thai-language job titles, skills, and occupations across large-scale labor market datasets
• Designed rule-based and statistical parsing systems using Python, SQL, and RegEx, improving classifier precision and recall by ~20%
• Addressed challenges in Thai text processing (tokenization, normalization, and ambiguity) to improve model accuracy and consistency
• Partnered with engineering, product, and linguistics teams to deliver scalable, production-ready language processing solutions

Cognex Corporation

Jan 2023 – June 2023

IT Business Applications Co-op

• Analyzed enterprise-scale Salesforce and SAP datasets to identify inefficiencies and recommend data-driven process improvements
• Translated business requirements into technical workflows, improving system usability and reporting clarity
• Led documentation, tracking, and delivery planning for global initiatives, improving execution efficiency and cross-team communication

Silicon Valley Bank

Jan 2022 – June 2022

Digital Marketing & Analytics Co-op

• Designed and analyzed A/B tests for email campaigns, improving click-through rates (CTR) by ~10–15%
• Built dashboards in Google Analytics to track conversion rates, engagement metrics, and user behavior
• Delivered actionable insights that informed data-driven marketing strategy and campaign optimization

projects

TasteMatch CA — Hybrid Recommender System (NLP + RAG)

• Built a hybrid restaurant recommender system combining collaborative filtering (ALS), content-based filtering (TF-IDF), and Retrieval-Augmented Generation (RAG)
• Developed NLP-based query parser to extract user intent (cuisine, price, location)
• Improved recommendation relevance by using feature engineering
• Evaluated performance using Precision@5 (0.82), NDCG (0.83), and MRR (0.85)
• Deployed an interactive Streamlit application for real-time recommendations

Python NLP RAG TF-IDF ALS ChromaDB Streamlit

View Code

MassWeatherHub — Real-Time Weather Data Dashboard

• Built a real-time data pipeline integrating multiple weather APIs (Open-Meteo, OpenWeather)
• Processed and merged geospatial datasets (TIGER/Line Shapefiles)
• Developed interactive dashboards and maps using Flask and Folium
• Improved data processing efficiency by ~30% through optimized transformations
• Enabled exploration of real-time and historical weather data

Python Flask REST APIs Folium Pandas Geospatial Data

Exoplanet Habitability Prediction (Machine Learning)

• Developed ML models (SVM, Random Forest, Logistic Regression) for habitability prediction
• Achieved F1-score up to ~0.78 through feature engineering and tuning
• Performed cross-validation and hyperparameter optimization
• Addressed class imbalance and multicollinearity
• Built visualizations to communicate insights

Python Scikit-learn Pandas NumPy SVM Random Forest EDA

skills

Core

Python
SQL (MySQL)
R
Machine Learning
Statistical Inference
A/B Testing

NLP / AI

Natural Language Processing
RegEx
NLTK
Embeddings
RAG
LLMs

Data & Visualization

Pandas
NumPy
Matplotlib
Seaborn
Tableau
EDA

Tools & Systems

PyTorch
Apache Spark
Flask
Streamlit
DBT
AWS
Git