I build machine-learning systems end to end, from scraping the data to serving the model behind an API.
BS Computer Science at Angeles University Foundation. Project lead on A.S.E.A., a transformer-based classifier that detects Philippine SMS scams written in code-switched Tagalog–English.
Generic spam filters miss Philippine fraud messages because they're written in code-switched Tagalog–English. A.S.E.A. classifies SMS into 12 scam-intent classes and shows why a message was flagged.
Owned model training, system deployment, and most of the technical documentation.
Coordinated dataset collection: a custom Philippine SMS corpus built with Scrapy and Tesseract OCR on message screenshots.
Fine-tuned and benchmarked six transformers against Naive Bayes, SVM, and LSTM baselines, with automated best-model selection.
Served the selected model through FastAPI, with a Streamlit + Firebase front end and SHAP explanations.
Co-authoring the research paper, targeting SOICT 2026.
Best model · mBERT · 12 classes
Outperformed the SVM, LSTM, and Naive Bayes baselines.
Accuracy
77.3%
Macro F1
0.73
AUC-ROC
0.96
Inference
8ms
Pipeline I built
01
Collect
Scrapy scraping + Tesseract OCR into a Philippine SMS corpus
02
Train
Six transformers vs. three classical baselines on Kaggle
03
Select
Best model picked automatically from comparison metrics
04
Explain
SHAP surfaces the tokens behind each verdict
05
Serve
FastAPI inference with a Streamlit + Firebase interface
Models benchmarked
mBERTselected
Tagalog RoBERTa
Tagalog BERT
Tagalog DistilBERT
Tagalog ELECTRA
ALBERT
Multinomial NB
SVM
LSTM
02 · More projects
Design, data, networks and the foundations underneath.
UI/UX Design · Figma
Parenting companion app: wireframe to mockup
A full design progression for a mobile app with a baby tracker, baby monitor, milk meter, calendar, diary, and notes. Structure and user flow were validated in low fidelity before any visual design.
Data Analysis · Power BI
Sales geography & purchase forecast
An interactive three-page report: customers by state and year, a daily purchase trend with a forecast band, and loyalty and spend summaries.
Networking · Cisco Packet Tracer
Topology design & LAN troubleshooting
Routed multi-segment topologies with IP addressing configured and connectivity verified end to end. Plus hands-on RJ45 termination on Cat5 UTP.
Database & Systems Design
Order management schema
Customers, orders, order items, products, suppliers, payments, deliveries, and inventory, modeled as an ERD, then mapped to keys and types and queried with SQL.
Python · Group final project
Caesar cipher encrypter
Character-level shifting with modular arithmetic for alphabet wraparound. Try it:
03 · Skills
Tools I've shipped with.
Python is my primary language. The ML and backend tools were used to build A.S.E.A.; the rest come from coursework and labs.
Programming & ML
Python
PyTorch
HuggingFace Transformers
SHAP
Kaggle Notebooks
Backend & data collection
FastAPI
Streamlit
Firebase
Scrapy
Tesseract OCR
Data & databases
SQL
ERD & normalization
Power BI
Python data analysis
Networking
Cisco Packet Tracer
TCP/IP addressing
LAN troubleshooting
RJ45 / Cat5 UTP
Design & documentation
Figma
Wireframing
Hi-fi prototyping
System flowcharts
Everyday tools
Git & GitHub
VS Code
MS Office
Hermes Agent
Hermes Desktop
Claude Code
Claude Desktop
Notion AI
Gemini
Christopher Pineda
04 · About
Curious first, then thorough.
I'm a Computer Science student at Angeles University Foundation. My path here didn't start in a traditional classroom. I earned my high-school equivalency through the Alternative Learning System, which taught me to learn independently and finish what I start.
On A.S.E.A. I was chosen as project lead and took ownership of the parts most likely to break: training, deployment, and documentation. I'm looking for an internship where I can contribute to real systems and learn how professional engineering teams build, review, and ship.