Software Engineer, Machine Learning
+82-10-3715-9048
Passionate engineer with 7 years of experience in the software industry, I am a team player and I enjoy tackling challenging problems.
Although I love Python and Machine Learning, I have interest in anything related to software development.
I'm looking for a collaborative team with a strong, positive impact on the world around us, to build great things together.
2022 ~ Now
NLP Tech Lead
Fleksy
Improved the quality of the NLP features of the Fleksy mobile keyboard : auto-correction, auto-completion, next-word prediction, swipe gesture recognition, in many languages (82+).
‣ Implemented automated NLP benchmark, to visualize how well our features evolve across languages and across versions
‣ Improved the Language Model leveraging open-source tools, enhancing auto- correction for 97% of the 82 languages, up to 28% better in some languages
2021 ~ 2022
Machine Learning Engineer / Engineering manager
Spacewalk
Developed AI for automated floor plan generation in this proptech startup, using Reinforcement Learning. Worked in a team of 8, managed 2 engineers.
‣ Implemented path checking algorithm with 87 times better performance than requirements
‣ Established best practices and documentation standards, adopted in every technical teams of the company
‣ Built a highly performant RL prototype (250 times faster than the existing model) which was later integrated into the main product, and used as guide for future work
2018 ~ 2021
Machine Learning Engineer
42Maru
I worked on several NLP tasks with a focus on text summarization within the research team of this Korean startup.
‣ SOTA in question generation with BERT architectures
‣ SOTA several times in text summarization
‣ Developed a framework for deploying Pytorch models to production easily, used across several teams
‣ Production-ready summarization API built from the ground up leading to a commercial product
Feb. ~ Jul. 2018
Intern in Quality Assurance
Gracenote - Nielsen
Developed automated tests at industrial scale for smart TV.
‣ Fully automated test suite for the ACR component, saving hours of manual tests for the team
‣ Improved deployment pipeline to a new quality standard
‣ Conducted tests for an admin dashboard, without supervision
Jul. ~ Aug. 2017