Joanito Agili Lopo

Joanito Agili Lopo

NLP Researcher

I work on multilingual and efficient large language models, with a focus on understanding and controlling language representations at inference time. My research asks how we can disentangle language and semantic information inside multilingual LLMs to enable robust cross-lingual transfer — especially for low-resource languages, including Indonesian local languages.

Multilingual NLP Low-Resource Languages Representation Learning

Research Interest

My primary interest is in how internal representations in multilingual LLMs encode language versus semantic information, and how this structure can be leveraged for efficient adaptation without large-scale retraining.

Recent work suggests language-specific and language-neutral information are partially separable in a model's latent space. I aim to formalise disentanglement as a structured objective and to build adaptive language control mechanisms at inference time.

Representation Analysis Probing & similarity across layers and model scales
Latent Disentanglement Separating language identity from semantic content
Inference-Time Control Adaptive language steering without retraining

Education

2025
Universitas Gadjah Mada
Master of Computer Science in Artificial Intelligence - NLP
Thesis: An Instructional Linguistic Approach For Training Continual Instruction Tuning-Based Translation Models In Kupang Malay Language
2023
Universitas Kristen Satya Wacana
Bachelor of Information System — Machine Learning
Thesis: Evaluating Sampling Techniques for Healthcare Insurance Fraud Detection in Imbalanced Dataset

Publications

2026
Improving Lexical Difficulty Prediction with Context-Aligned Contrastive Learning and Ridge Ensembling
Muhamad et al. · BEA 2026 Shared Task
2026
Normalization and Deterministic Parsing for Formal Reasoning in LLMs
Muhamad et al. · SemEval-2026 Task 11
2025
Language Surgery in Multilingual Large Language Models
Lopo et al. · MRL 2025 (5th Workshop on Multilingual Representation Learning)
2025
An Instructional Linguistic Approach For Training Continual Instruction Tuning-Based Translation Models In Kupang Malay Language
Lopo et al. · Master Thesis
2025
Reconstructing Autoencoder-like MLP Architecture for Feature-Preserving Pooling in CNNs
Lopo et al. · ICoAIT 2025 · Best Paper Award
2024
CIKMar: A Dual-Encoder Approach to Prompt-Based Reranking in Educational Dialogue Systems
Lopo et al. · PACLIC 2024 (38th Pacific Asia Conference on Language, Information and Computation)
2024
SEACrowd: A Multilingual Multimodal Data Hub and Benchmark Suite for Southeast Asian Languages
Lovenia et al. · EMNLP 2024
2024
Constructing and Expanding Low-Resource and Underrepresented Parallel Datasets for Indonesian Local Languages
Lopo & Tanone · arXiv preprint
2023
Enhancing Multi-Output Time Series Forecasting with Encoder-Decoder Networks
Hartomo, Lopo & Purnomo · JISEBI, 9(2), 195–213
2023
Evaluating Sampling Techniques for Healthcare Insurance Fraud Detection in Imbalanced Dataset
Lopo & Hartomo · JITEKI, 9(2), 223–238

Projects

Bhinneka Korpus ↗
Parallel corpus for Indonesian regional languages with linguistic annotations, statistical MT models, and a public translation playground.
NLP · Low-resource · 2023
T5 Abstractive Summarisation
Fine-tuned T5 on the Liputan6 Indonesian news dataset for concise abstractive summarisation of long-form articles.
NLP · PyTorch · 2022
Sentiment Analysis
Benchmarked four ML algorithms (LR, Logistic, Naïve Bayes, SVM) on Instagram comment sentiment classification.
ML · Scikit-learn · 2023
i3gis CLI Tool
Command-line interface in Go integrated into the i3gis DevSecOps platform, developed with Agile practices.
Go · DevSecOps · 2022
Customer Segmentation
Clustering-based segmentation to analyse customer behaviour patterns in a marketing campaign dataset.
ML · Clustering · 2021

Experience

Apr 2026 – July 2026
UGM, Yogyakarta
Research Assistant
Conducted AI research, assisted paper revisions, and supported research implementation and publication preparation.
Sep – Dec 2024
UGM, Yogyakarta
Computer Lab Assistant
Supported students with C++ programming, debugging, and project development; maintained lab systems.
Jan – Jun 2023
UKSW, Salatiga
Research Intern
Designed ML pipelines for Parkinson's detection, fraud detection, multi-output time series forecasting, and landslide prediction.
Jul 2022 – Jan 2023
PT Inovasi Informatika Indonesia
Software Developer Intern
Built a Go CLI application for the i3gis DevSecOps platform; led development and production cycles using Agile.
Jul – Aug 2021
Kopernik, Soe NTT
Enumerator
Conducted interviews with mothers and led focus group discussions on child stunting for a public health research programme.

Awards & Honors

Contact

✉ amalopo99@gmail.com ⌥ github.com/joanitolopo ↓ Download CV