Apr 2020 – May 2022 · 2 yrs 1 mo · Suwon, Gyeonggi, South Korea

• Leveraging the power of Samsung’s proprietary data spread across multiple touchpoints(TV, OTT, Mobile) to deliver value for advertisers and consumers.
• Technical Product Management for Samsung’s Big Data Platform and Data Warehousing. Helping build a consolidated data platform to gather data from diverse Samsung’s services and enable targeting at scale
• Enabling efforts to integrate data from 1st party proprietary ACR(Automatic Content Recognition), 3rd party data, and mobile usage with a special focus on privacy
• Working with cross-functional teams in Engineering, Product Marketing, Privacy, and UX spanning across US, KR, CA
• Leading efforts with PMMs to prioritise usage of proprietary mobile data for effective targeting and retargeting
• Others: DSP, SSP, Ad-exchanges, Publisher Platform, Ad Ratings, Attribution, CTV Ads Management, Global expansion

Sep 2018 – Apr 2020 · 1 yr 7 mos · Suwon, Gyeonggi-do, Korea

• Machine Learning Engineer with a focus towards improving the picture quality on Samsung TVs. Worked towards multiple projects including:
• Enhancing the 'Ambient Mode' of Frame TVs - Performing 'Style Transfer' for user-uploaded images on-device. Using Generative Models for creating new artworks to be shown on Frame devices. Automatic transformations on images based on the time of the day.
• Frame Interpolation using Deep Learning based solutions, to replace the incoming 30fps videos into higher frame rates in real-time, with support up to 4K videos.
• Majority contributor to 1 patent (KR filed, US filing pending)

Jan 2018 – Jul 2018 · 6 mos · Greater Seattle Area

• Worked on NLP-based research work to be used within Amazon services to its customers.
• Developed multi-sense embeddings for words which assign multiple vector representations to words in order to improve the 'single-sense' word embeddings. We generate our embeddings using information from topic modeling which is a mixed membership model that assigns a topic distribution to each document. The project involved formulating the problem mathematically and subsequently developing the model architecture.
• https://arxiv.org/abs/1909.07746

May 2017 – Jul 2017 · 2 mos · Suwon, Gyeonggi-do, Korea

• Worked on Demographic Prediction of households for targeted advertisement to customers, content recommendation, user profiling
• Worked on ETL and data preparation for various types of models which were used
• Developed deep learning models such as CNNs, RNNs, Adverserial models, TOL for household based time-series/non-time based profiling of customers
• Researched on various ensemble models and implemented them for improving the predictions
• Tried and tested various model configurations and obtained a good research perspective

May 2016 – Jul 2016 · 2 mos · Noida Area, India

• I worked on two projects. The first one related to image matching of images clicked by phone camera with thousands of images in the database. It involved extensive use of opencv library. The second one involved working on automatic text classification for a leading blog publisher. It also involved a client meeting in Singapore.