I like machine learning, working with data,
and building automations.
Drag things around.
On desktop, drag to rearrange. With a move button focused, use arrow keys, Shift for larger steps, or Home to restore its position.
Text transcripts become summaries through a fine-tuned Pegasus model. The project includes notebooks exploring T5, LED, BART, and Pegasus.
Transcript → Pegasus → Summary
An Ames Housing regression project connecting model training to a Streamlit prediction interface, with Random Forest and XGBoost experimentation.
Ames Housing · Streamlit interface
Grayscale image representations are classified using 60 GLCM texture features. Training scripts compare Random Forest, SVM, and XGBoost.
Image representations · Flask interface
Worked on content review ranking in Python and scikit-learn, ranked action queues with reason codes and confidence labels, and search-performance analysis with DuckDB SQL.
Built a product recommendation chatbot and automated Bigin CRM updates from voice notes and text messages. Created case and new business opportunity trackers, supported by n8n workflows, MySQL, and Supabase.
Python · C · JavaScript
HTML · CSS
scikit-learn · PyTorch · TensorFlow
NLTK · LangChain · RAG · Embeddings
SQL · pandas · DuckDB
MySQL · Supabase
n8n · Prompt engineering
CRM workflows · Model evaluation
I’m Finley Septhian Darma, a Computer Science undergraduate at Bina Nusantara University in Indonesia.
I’m especially interested in experimenting with machine learning models and connecting tools through automation.
I also enjoy helping others learn. I’m a SASC scholarship tutor and a former BNCC Learning and Training Activist, where I contributed to the programming curriculum, exercises, and learning materials.