Monash University · Melbourne, Australia
Portfolio · LinkedIn · ORCID · Kaggle
I am pursuing a Master of Artificial Intelligence at Monash University, researching generative AI and computer vision. I work on text-to-image diffusion models, efficient adaptation, and evaluation.
Previously, I worked as a Data Analyst at Indonesia’s Ministry of Finance, developing machine learning solutions, analytics dashboards, web applications, and digital systems.
Research interests: Generative AI · Computer vision & multimodal learning · LoRA & hypernetworks · AI benchmarking
Project notes, results & research links
- Synthetic Image Attribution: Our 1st-place ICANN 2026 DLMMDD solution combines a five-fold EVA02-Large ensemble, temperature calibration, and test-time augmentation. The private-best submissions achieved 99.73% accuracy. We study marginal balancing using the challenge’s known class composition. Competition.
- Cryptanalysis Benchmark: I built 11 story-driven tasks across classical cipher families. Models extract cipher rules from narrative clues and recover exact plaintext, with deterministic scoring and a growing public leaderboard.
- MuseumSCAT: My 3rd-place solution combines vision-language transcription, confidence ranking, and selective, evidence-supported corrections to specimen dates and localities. The selected submission achieved a private AURC of 0.00554. Reproduction guide.
- CodelessML: A Streamlit application for tabular data exploration, classification or regression model comparison, model export, and prediction workflows. Users upload CSV or Excel data and compare nine algorithms per task. Journal article (2024).
| Year | Result | Work |
|---|---|---|
| 2026 | 1st place | ICANN DLMMDD Synthetic Image Attribution Challenge |
| 2026 | 3rd place | MuseumSCAT Specimen Collection Annotation Task |
| 2026 | 3rd place | Forams2026: detection and classification in micro-CT volumes |
| 2025 | 2nd place | Advent of MAPS 2025 Programming Challenge |
| 2024 | 3rd place | GovAI Hackathon: satellite imagery and carbon estimation |
Programming & data: Python, SQL, pandas, NumPy.
Machine learning: PyTorch, scikit-learn, XGBoost, CatBoost.
Generative AI: diffusion models, Transformers, LoRA, Hugging Face.
Development & research: Git, Linux, Jupyter, Streamlit, LaTeX.
GitHub activity & public-code languages
Language proportions describe public repository code, including notebooks; they do not measure proficiency. These optional cards are provided by GitHub Readme Stats.
I welcome conversations about generative AI, computer vision, personalisation, and applied machine learning. Connect on LinkedIn or explore my portfolio.