About

I’m currently a machine learning developer at SAP’s innovation center in Singapore, where I research and develop ML-based solutions for enterprise use cases. I graduated from Yale-NUS College in 2017 as part of its inaugural cohort.

Full CV

profile

Experience

SAP (July 2017 - July 2020)

Machine Learning Developer. Currently working on SAP Cash Application, SAP’s flagship machine learning solution that automates manual accounting procedures using machine learning algorithms. I was the youngest and first bachelor hire for the position.

Code Gakko (May 2016 - present)

Founder. A coding academy for students of all ages, currently operating in Singapore and Manila.

The Tor Project (May 2015 - June 2016)

Developer. Worked with Virgil Griffith on Roster under the Open Technology Fund Emerging Technologies Fellowship.

STYLER (June 2015 - May 2016)

Developer. Used Neo4j and other web technologies to build a graphical database that visualizes and analyzes user behavior and interactions on the company’s app.


Projects

Contextual AI

SAP’s first open-source machine learning library for explainability. It aims to address the trust gap between machine learning systems and their users. Check out the release blog post here.

Electionaire

An online questionnaire that matches users with one of ten Singaporean political parties. Created to provide insight to the electorate of Singapore before its 2015 General Election. Launched one week prior to the elections, the app received 670,000+ page views and over 180,000 unique users.

Media mentions:

Roster

A now-deprecated Tor project that measures various statistics about the Tor network and its relays, including up-time, bandwidth, and geographical diversity of relay families. It awards points and badges to relay operators based on the performance of these families.


Publications and Preprints

Sean Saito, Eugene Chua, Nicholas Capel, Rocco Hu. Improving LIME Robustness with Smarter Locality Sampling. 2nd Workshop on Adversarial Learning Methods for Machine Learning and Data Mining @ KDD 2020 (2020). (link)

Sean Saito, and Jin Wang. Explaining Away Attacks Against Neural Networks. First Workshop on Secure and Resilient Autonomy, Third Conference on Machine Learning and Systems (MLSys) (2020). (link)

Sean Saito, and Sujoy Roy. Effects of Loss Functions And Target Representations on Adversarial Robustness. arXiv preprint arXiv:1812.00181 (2018). (link)

Sean Saito, Yang Wenzhuo, and Rajalingappaa Shanmugamani. Python Reinforcement Learning Projects: Eight Hands-on Projects Exploring Reinforcement Learning Algorithms Using TensorFlow. Packt Publishing. September 29, 2018. (link)

Wu, Zhenzhou, and Sean Saito. HiNet: Hierarchical Classification with Neural Network. arXiv preprint arXiv:1705.11105 (2017). (link)


Patents

Sean Saito, Chaitanya Krishna Joshi, Rajalingappaa Shanmugamani, Truc Viet Le, and Rajesh Vellore Arumugam. “Utilizing embeddings for efficient matching of entities.” U.S. Patent Application 16/217,148, filed June 18, 2020.

LE, Truc Viet, Sean Saito, Chaitanya Krishna Joshi, and Rajalingappaa Shanmugamani. “GRAPHICAL APPROACH TO MULTI-MATCHING.” U.S. Patent Application 16/210,070, filed June 11, 2020.

Sean Saito, Truc Viet Le, Chaitanya Joshi, and Rajalingappaa Shanmugamani. “Representing sets of entities for matching problems.” U.S. Patent Application 16/208,681, filed June 4, 2020.

Sean Saito, and Sujoy Roy. “Platform for preventing adversarial attacks on image-based machine learning models.” U.S. Patent Application 16/186,784, filed May 14, 2020.


Hackathons

Data Science Game 2016 - Finalist (7th out of 143 teams)

The largest student data science competition sponsored by companies like Microsoft, Capgemini, and AXA. Competed among a group of graduate students. We also talked about our experience at a data science meet-up in Singapore.

Battlehack Singapore 2015 - First runner-up

Created Portal, an Android application that harnesses beacon technology to enhance marketing. When a user is interested in a product on a poster ad in a public space, he or she can receive a virtual version of that ad on a smartphone and purchase the product directly from the app.

Emirates Travel Hackathon 2015 - First runner-up

Created Aviato, an online travel platform that aggregates data from travelers’ social network profiles including Facebook to make better recommendations to customers, while providing deeper insight on travel trends for companies.

A*STAR IoT Hackathon 2015 - Winner

Created Blackhat, a blackbox + construction helmet. Attaching various sensors, including a thermometer, a hygrometer, and an IMU to a construction helmet, we created a wearable that can record data about the immediate surroundings of each construction worker, providing insight about the safety conditions of the work environment.


In the News

Aiming for AI, machine learning jobs after school? Don’t just focus on tech, grads say

The undergrad who started a coding academy


🔵 🔴