Adrianus Saga Ekakristi
Location: Jakarta, Indonesia
I graduated from Universitas Indonesia with Master's degree in Computer Science. My primary research interest is in natural language processing (NLP), while my career experience is in big data processing.

At work, my team and I are responsible for developing, maintaining, and innovating both stream and ETL data processing pipeline. I work with scalable and distributed technology stacks, including Hadoop, Kafka, Kubernetes, Cassandra, GCP Dataflow, Dataproc, BigQuery, Azure Databricks, and many more.

My research interests include natural language processing, information retrieval, and transfer learning. In the future, I am interested in exploring interpretable and explainable AI as well as neuro-symbolic paradigm.
Experience
Telkom Indonesia
Jakarta, Indonesia
December 2021 - November 2022 (Full-Time)
May 2024 - March 2025 (Part-Time)
Data Engineer
Responsible for developing end-to-end data platform with lambda architecture that consists of data ingestion and change data capture (CDC), stream and batch data pipelines, scheduling platform, and query engine. We use several cloud technology such as GCP Pub/Sub, Dataflow, GCS, GKE, and BigQuery. I am also responsible for improving data acquisition and data processing for a government project.
Bukalapak
Jakarta, Indonesia
September 2017 - December 2021
Data Engineer
Responsible for developing, maintaining, monitoring, and innovating our Extract-Transfom-Load (ETL) and stream data processing. Mainly, we are working with big data platform and framework, such as HDFS, Yarn, HBase, Spark, Hive, Kafka, Presto, Cassandra, Airflow, along with Kubernetes, Docker, and many other tools. We are also using several cloud services, such as GCP Dataproc and BigQuery.
Kios Untuk Dagang Online (KUDO)
Jakarta, Indonesia
June 2016 - August 2016
Back-End API Developer Intern
Building application programming (API) interface for Kudo Field Officer Application (FO App) project. FO App is used by Kudo Field Officers to support their job, which are overseeing and supporting Kudo Agents. The API provide all data required for mobile application, web application, and content management system.
Education
Universitas Indonesia
Depok, Indonesia
January 2023 - January 2025
Master
Computer Science
Universitas Indonesia
Depok, Indonesia
July 2013 - July 2017
Undergraduate
Computer Science
Kolese Kanisius Jakarta
Jakarta, Indonesia
2010 - 2013
Senior High School
Science Stream
Research Experience
Finding Questions in Medical Forum Posts Using Sequence Labeling Approach
We propose a sequence labeling method using Conditional Random Fields for question processing. This paper was presented at CICLing 2018. Link: Springer
Corpus Development for Indonesian Consumer-Health Question Answering System
We present a medical question answering corpus consists of question-answer pairs from five Indonesian medical forums. This paper was presented at ICACSIS 2017. Link: IEEE
Medical Question Text Decomposition with Sequence Labeling and Chunking Approach
In this undergraduate thesis, we propose a pipeline for decomposing informal, complex health consultation request text in medical question answering system. Link: CSUI Library
Teaching Experience
Teaching Assistant Coordinator for Information Retrieval
September 2023 - December 2023
Universitas Indonesia
Depok, Indonesia
Teaching Assistant for Programming Foundation
September 2016 - December 2016
Universitas Indonesia
Depok, Indonesia
Teaching Assistant for Declarative Programming Course
January 2016 - June 2016
Universitas Indonesia
Depok, Indonesia
Skills
Programming Language:
Python, Java, Scala, Go, PHP, Assembly, F#
Open-Source Data Technology:
Hadoop, YARN, HBase, Spark, Kafka, Hive, Presto, Ignite, Delta Lake, MinIO, Dask, Ray
Cloud Data Platform:
GCP Dataflow, Dataproc, GCS, CloudSQL, BigQuery, Azure Databricks
AI Tools:
PyTorch, Horovod, Tensorflow, Weka, NLTK, Keras
Provisioning Tools:
Ansible, Terraform
Distributed Orchestration Framework:
Kubernetes, Helm Charts, Mesos, DC/OS
Database:
Cassandra, MySQL, PostgreSQL, MongoDB, Qdrant
Web Framework:
Flask, Laravel, Yii, Django
Version Control System:
Git, SVN
Others:
Airflow, Prometheus, Grafana
Side Projects
Neural Symbolic Approach with Generative Model for Machine Reading Comprehension
In this research project, a neural symbolic approach with an executable symbolic program generator was proposed to answer questions that require discrete reasoning. The T5 model and its variants were used in the experiment to answer the DROP dataset. This work was done as the final project for the Advanced Information Retrieval course.
Forest Wildfire Severity Estimation from Satellite Imagery with Multi-Task Learning Approach
In this research, multi-task learning framework is proposed using land cover segmentation as auxiliary task for estimating forest wildfire severity. This experiment involve various UNet and Visual Transformer models. This research was done as final project for Advanced Spatial Data Analysis course.
Flow-based Network Traffic Classification using Multi-Task Learning Approach
In this project, multi-task learning approach is proposed to classify network traffic class using Hierarchical-CNN and LSTM-CNN model architecture for extended QUIC dataset. This experiment was done as final project for Advanced Computer Networks course.
Multi-Task Learning for Indonesian Part-of-Speech Tagging and Named Entity Recognition
In this research project, we explore various architecture of multi-task learning approach for Indonesian Named Entity Recognition task with Part-of-Speech Tagging as an auxiliary task using pre-trained BERT model. This work was done as final project for Research Methodology course.
Proyeku
A website-based platform to facilitate freelancer, project owner and freelancer seeker. This website was implemented using PHP Laravel framework and was built for main project assignment of Software Project course.
Wireless Camera and Video Recorder built using Raspberry Pi
Photo taking app and video recorder built with Raspberry Pi model B, a standard USB webcam, a wireless adapter, and LED. The features were implemented in Raspbian Wheezy OS. This device was built as final project assignment of System Programming course.
2D Typing Game with Unity Engine: Typing Wizard
2D typing game for Windows PC platform. This game was built using Unity engine. This game was built as the final project assignment of Game Development course at Fasilkom UI and can be accessed at Game Jolt.
Interests
Data Engineering, Data Architecture, Natural Language Processing, Information Retrieval, Transfer Learning, Interpretable and Explainable AI, Neuro-Symbolic AI.