Experience

Machine Learning Engineer at PropertyGuru - Data Science Center of Excellence, Singapore
June 2022 - present

• Developed an Automatic Valuation Model that analyzes historical sales transactions, property characteristics, and price trends to estimate the current market value for a specific property. Contributed to the company’s revenue by productionizing the solution as a sellable system to external clients, including banks and real estate consultants.
• Developed an Address Matching system to synchronize a physical location with address records in the data warehouse. Leveraged advance natural language processing techniques to aggregate data from multiple sources to provide rich information from an address, e.g. property information, property valuation, and property risk assessment. Improved the matching rate by more than 70% compared to the legacy SQL-based string matching.
• Led cross-team collaboration with data engineers and software engineers to solve inaccurate address listings in Malaysia. Fixed 150k inaccurate listing data, resulting in more accurate location-based listing recommendations and an estimated six-month revenue increase of over USD100k.
• Led an initiative to incorporate Trustworthy metrics in the machine learning systems. Detected and performed root cause analysis on unacceptable fluctuating predictions in a highly critical product. As a result, the volatile predictions were reduced by more than 90%.

AI Research Engineer at SCIS SMU, Singapore
Nov 2019 - May 2022

• Conducted research on Trustworthy AI supervised by ACM Distinguished Scientist Prof David Lo.
• Published 7 papers in international conferences and journals, i.e., IEEE TSE (Q1, CCF Rank A, Impact Factor: 6.11), ACM ESEC/FSE (CORE A*), IEEE ICSME (CORE A), and IEEE SANER (CORE A).
• Experienced in maintaining various TensorFlow and PyTorch machine learning projects using Docker running on Nvidia DGX Station AI Supercomputer.

Co-Founder & CTO at Alinamed, Indonesia
Jan 2019 - July 2020

• Managed the end-to-end software development lifecycle by collaborating across members to evolve SDLC, including functional requirements analysis, design specification, technical code review, project planning, and implementation of applications
• Managed the selection of programming languages, frameworks, cloud, web servers, databases, and mobile apps
• Won prestigious startup competitions, e.g., 1st Winner at WMM by Bank Mandiri, 1st Winner at Startup Weekend Surabaya, Winner at Tech-to-Impact by UK-ID Tech Hub. Secured IDR 750 million investment from Telkom Indonesia.

Machine Learning Engineer at Bukalapak, Indonesia
May 2018 - Aug 2018 (Internship)

• Developed object detection as a part of the “search by image” feature. It’s implemented using Faster RCNN on Tensorflow
• Developed character recognition to detect seller’s information (e.g., phone number and Instagram username) in the e-commerce product image. Localized text lines using CTPN (Connectionist Text Proposal Network). Implemented post-processing techniques on OpenCV to combine several words within the same context.

Honors and Awards

• 2022 | Bronze Medal - Top 7% (170 from 2,654) | Evaluating Written Essays | NLP + CV | Kaggle Leaderboard
  The challenge was to assess the language proficiency of students by analyzing their written essays. I leveraged DeBERTa language models to extract the feature embeddings from the essays. Then, I trained neural network layers to predict the proficiency scores from extracted features
• 2021 | Silver Medal - Top 5% (101 from 2,426) | Shopee Product Matching | NLP + CV | Kaggle Leaderboard
  The task was predicting similar e-commerce products by using product images and titles. I utilized EfficientNet and NFNet to extract feature embeddings from images and utilized BERT and TFIDF to extract feature embeddings from texts. I calculated the cosine distance between products and employed k-NN to find similar products
• 2021 | Top 3% (28 from 1,034) | Information Extraction - Shopee Data Science Challenge | NLP | Kaggle Leaderboard
  The challenge was extracting Point of Interest (POI) names and street names from unformatted Indonesia addresses collected by Shopee. I framed the challenge as NER (Named Entity Recognition), where POI and street are the extracted entities. Then, I fine-tuned the IndoNLU model on the customized dataset.
• 2019 | Gojek Upscale 02 for Indonesia’s Data Science Enthusiast | Certificate
  A complete two-day experience from the Gojek data science team. It covered machine learning, data modeling, and Gojek data science practices. I was one of 20 chosen participants out of 1,000 applicants
• 2019 | Southeast Asia Machine Learning School (SEA MLS)
  A five-day ML training event held by Southeast Asia’s AI experts working at DeepMind UK and Google Brain USA. I was one of 200 chosen participants out of 1200 applicants across Southeast Asia and India
• 2017 | 1^st Winner | Competition on Building Device for Visually Impaired People | Documentation
  Developed a device for visually impaired people to help them hear what they see. It’s powered by a RaspberryPI as a processing unit. It captured an image using a camera, recognized texts and then output voices
• 2011 | Bronze Medal | Mathematics Olympiad at 2nd National Science Camp | Certificate
  Awarded by Ministry of Education - Indonesia

Technical Skills

• Programming: Python, C++, JavaScript, Java, C, PHP, HTML, CSS, React Native, Golang,
• ML Library: PyTorch, TensorFlow, Keras, OpenCV, Scikit-Learn
• Cloud: AWS, GCP, Microsoft Azure
• Database: MySQL, PostgreSQL, MongoDB

Education

B.Eng. in Computer Science, Institut Teknologi Bandung, 2019

• Cum Laude with GPA 3.68 / 4.00
• I am proud to have earned the distinction of being recognized as an inspiring graduate, a testament to my exceptional record of academic excellence and extracurricular accomplishments

Certifications

• AWS Certified Solution Architect - Associate
  by Amazon Web Services | Credential
• Improving Deep Neural Networks: Hyperparameter Tuning, Regularization and Optimization
  by DeepLearning.ai | Credential
• Neural Networks and Deep Learning
  by DeepLearning.ai | Credential