CV

Bachelor's Degree

Universitas Indonesia

Computer Science

• Data Structure & Algorithm, Algorithm analysis
• Machine Reinforcement Learning, Deep Learning
• Statistic and Probability, Information Retrieval
• Linear Algebra, Calculus, Discrete Mathematics

IWSLT-2025 Low-Resource Languages Speech Translation Technology: Maltese to English

Developed a Maltese-to-English speech translation system. Fine-tuned OpenAI's Whisper V3 Large and implemented joint learning. Utilized QLoRA to optimize performance while maintaining computational efficiency.

SemEval-2025 Task 11: Evaluating State-of-the-Art Encoder for Multi-Label Emotion Detection

Conducted a systematic comparison of current state-of-the-art models for multilingual multi-label emotion detection, evaluating various encoders and training strategies. Highlighted the superiority of prompt-based encoders like BGE and mE5 combined with tree-based models over fully fine-tuned transformers. [On Submission]

Crowdsource, Crawl, or Generate? Creating SEA-VL, a Multicultural Vision-Language Dataset for Southeast Asia

Open-source dataset creation for SEA languages, contributing 300 data points. On submission to ACL 2025.

ChatGPT Assistance on Biochemistry Learning Outcomes of Pre-Service Teachers

Performed A/B testing to find out whether there is a statistically significant difference between accuracy and time taken for ChatGPT-assisted teaching versus non-ChatGPT-assisted teaching. Paper will be published soon.

Investigating the Challenges of First-Year Education Students in Basic Physics I Practicum

Performed A/B testing and experiment design to analyze the challenges of lab practicum sessions. Accepted in Journal of Research and Physics Education Studies (JRKPF).

Beyond the Gap: Understanding Gender Disparities in Indonesia’s National Science Olympiad Achievements

Conducted a mixed-method research investigating gender disparities in Indonesia's National Science Olympiad (OSN) achievements. The research found significant male dominance across subjects and explored the impact of social stereotypes and biological factors on female performance. Accepted in Daengku, Journal of Humanities and Social Sciences Innovation 2024.

Exploring Indonesian Consumer Health Question Answering using Active Multi Retrieval

Optimizing integration of RAG and LLM in Indonesian language assisted by ColBERT retrieval to reduce hallucination and enhance confidence using an Active Multi Retrieval System.

2nd Winner - Gemastik Data Mining

Ministry of Education, Culture, Research, and Technology

Advanced through preliminary rounds with a research paper titled 'Exploring Indonesiaon Consumer Health Question Answering utilizing Active Multi Retrieval', dominated the leaderboard in ordinal regression for university tuition cost optimization.

Silver Medal - Indonesian National Olympiad in Informatics

Ministry of Education, Culture, Research, and Technology

Secured 8th place in Indonesia's biggest high school competition, earning a direct admission to Universitas Indonesia without any test.

Finalist - Smart Logistics Datathon

The Chinese University of Hong Kong

Competed against Asia's best universities in a 36-hour datathon addressing Hong Kong logistics challenges with theoretical and computational solutions.

Rank 15 - ICPC Regional Asia Jakarta

International Collegiate Programming Contest (ICPC)

Solved 13 algorithmic and mathematical problems under time pressure representing university-level programming excellence.

Finalist - ICPC Regional Asia Jakarta

International Collegiate Programming Contest (ICPC)

Competed in world-level algorithmic and mathematical problems in one of Asia's most prestigious programming competitions.

1st Winner - InnovationQuest

Airlangga University

Developed 'Road Damage Level Detection' using multimodal text and image fusion, combining YOLO, ResNet, BERT, and CatBoost.

3rd Winner - Dataquest

Airlangga University

Developed time series rainfall prediction and solved an OCR task related to medicine, securing 2nd in prelims and 3rd in finals.

Semi-finalist - Satria Data Competition

Ministry of Education, Culture, Research, and Technology

Built Indonesian NLI models trained on translated datasets using distributed cloud training to predict legal harmony in constitution documents.

2nd Winner - Datavidia

Bandung Institute of Technology

Developed a hybrid FCN-LSTM rainfall prediction model and designed an efficient earthquake early warning system.

3rd Winner - ILPC (Informatics Logical Programming Competition)

UBAYA University

Excelled in logic and discrete mathematics challenges and teamwork-based coding games in final stage.

5th Winner - Informatics Olympiad (IO)

Jember University

Solved theoretical computer science problems covering discrete mathematics, logic, and graph theory.

Contextual Modeling: Racism in Text Identification

Natural language processing model that predicts whether a given reply to a given context is racist. Research is mainly focused on developing a model that could understand context. Results show our technique could use context in some cases using a certain representation.

Optical Character Recognition for Medicine

Computer Vision approach to obtain medicine’s name from an arbitrary perspective of a medicine. Utilizing image processing to create a deep learning model pipeline that generalizes better over perspective, rotation, warping, and color to adapt better on new image data points.

Indo NLI XLMR Base Fine-tune

Natural Language Inference (NLI) entails determining if two statements contradict, are neutral, or entail each other. The XLMR Base model was fine-tuned using a vast dataset of 4 million translated texts. To optimize training on this large dataset, employing multiple techniques, including data parallelization, gradient accumulation, gradient checkpointing, and NVIDIA mixed precision is a must. Trained on 2 RTX 6000 GPUs on cloud computer. Outperforms others on key datasets.

Personal Stock Analytics

Using LSTM to analyze price changes. Combine with sentiment analytics scraped from social media. The model predicts a confidence interval. When the condition is not predictable, the predicted range will be really big. Not deployable model because of lack of generalization and predictive features.

Instagram Scraper Module

Robust Multiprocessing module to scrape massive amounts of Instagram data. Scrape post, profile picture, following, follower, and create a graph representation of a friendship. Focus mainly on finding hidden API and bypassing any required rendering in order to achieve full speed.

Data Scientist

Data Science Initiative, Youth Catalyst Foundation

Working directly under CTO to solve business problems by translating them into technical data objectives. Built a complete data infrastructure campaign using Mixpanel. Developed recommendation systems and a RAG chatbot for suggesting competitions, mentors, internships, scholarships, volunteer opportunities, and events.

Expert

Kaggle

Actively participated in discussions and published notebooks that garnered upvotes, earning a place in the top 3,300 out of 325,000 entries. Participated in 25 competitions, including a public playground competition ranked 23rd out of 934.

Member

Ikatan Alumni Tim Olimpiade Komputer Indonesia

Prestigious Indonesian olympiad winner organization regularly holding task force for national olympiad, national training, and an annual meetup connecting Indonesia's top computer science talents.

Head of Organizers

RISTEK x Datacamp Scholarship

Managed the distribution of 50 Datacamp premium scholarships and ensured a high-quality learning experience.

Teaching Assistant

RISTEK Sister in Tech

Mentored women in tech, provided materials and project tasks to enhance their portfolios.

Lead of Data Science & Analytics

RISTEK

Led the Data Science & Analytics team, organized Indonesia’s first Ristek Datathon, and developed programs supporting AI transformation in Indonesia.

Expert Staff of Data Science Academy

COMPFEST

Developed a 12-day data science curriculum for 12 speakers during the bootcamp. Successfully invited the CEO of a leading AI company in Indonesia.