Drug review sentimentFIG. 01
RECORDED NOTEBOOK RESULTS Reading between the ratings.
Comparing seven NLP models to understand positive, negative, and mixed drug reviews.
Read the case study DATA SCIENCE · AI · MARKET RESEARCHPORTFOLIO / 2026
Sebastian Latiphy.
Finding the signal.
Making it matter.
of data, intelligence, and people.
++
01 | RAW OBSERVATIONSn = 192
A LITTLE CURIOSITY. A LOT OF POSSIBILITIES.
01 / THE PORTFOLIO7 PROJECTS, DIFFERENT PERSPECTIVES
Curiosity, applied.
Research and working applications, explored through data and evaluation.
Showing 7 of 7 projects
Facial expression classificationFIG. 02
RECORDED NOTEBOOK RESULTS Small images. Bigger questions.
Testing a custom residual CNN against classical models and transfer learning on FER2013.
Read the case studyCredit-card classificationFIG. 03
RECORDED NOTEBOOK RESULTS Almost perfect. Look closer.
Why near-perfect accuracy can hide minority-class errors in an imbalanced dataset.
Read the case studyBusiness InsightsFIG. 04
APPLICATION ARCHITECTURE From numbers to next steps.
A sales dashboard with shared filters, interactive charts, and AI analysis grounded in selected data.
Read the case studyMakefitsFIG. 05
WARDROBE WORKFLOW / ILLUSTRATION A wardrobe, with a system.
A personal wardrobe app built to organize clothes, compose outfits, and plan what to wear.
Read the case studyBanking AI Risk LabFIG. 06
EVALUATION WORKFLOW / PROTOTYPE Trust, examined.
An inspectable policy assistant with retrieval controls and a risk evaluation workflow.
Read the case studyPharma demand forecastingFIG. 07
RECORDED NOTEBOOK RESULTS Demand changes. So does the horizon.
Exploring pharmaceutical sales forecasts and how the prediction horizon changes the model comparison.
Read the case study02 / BEHIND THE WORK
A curious mind.
An analytical lens.
I’m Sebastian. My interests sit where data science, artificial intelligence, and market research meet. This is a space for exploring questions, making sense of complexity, and sharing what I learn.