Ujjawal Soni

AI Researcher

Bengaluru, Karnataka, India8 yrs 3 mos experience

Most Likely To SwitchHighly Stable

Key Highlights

Expert in deep learning and computer vision solutions.
Developed real-time AI solutions for flagship smartphones.
Published research on fairness in recommendation systems.

Stackforce AI infers this person is a Data Science and Computer Vision expert in Consumer Electronics and Tech Research.

Contact

ujjawal9180@gmail.com LinkedIn

Skills

Core Skills

Data ScienceMachine LearningComputer VisionDeep Learning

Other Skills

AlgorithmsCC++Data AnalyticsData MiningData StructuresFairness in RecommendationsInstance SegmentationJavaKerasMatlabNatural Language Processing (NLP)Object DetectionPersonalizationPyTorch

About

I am currently working in the Data Science team at CRED, primarily focusing on personalisation and recommendation. Previously, I was working as a Lead Engineer (Machine Learning) at Samsung R&D Institute Bengaluru, India. As part of the Visual Understanding team, I worked on developing efficient computer vision solutions to solve complex machine learning problems in real-time and on-device. Prior to the corporate world, I graduated from the Indian Institute of Technology Madras with B.Tech. and M.Tech. in Computer Science & Engineering.

Experience

Cred

Data Science

Aug 2021 – Present · 4 yrs 7 mos · Bengaluru, Karnataka, India

Data SciencePersonalizationRecommendationMachine Learning

Samsung r&d institute india - bangalore private limited

2 roles

Lead Engineer

Promoted

Mar 2020 – Jul 2021 · 1 yr 4 mos

As a member of the Visual Understanding team:
Designed, trained, and deployed efficient deep neural network architecture for object detection in images. Developed solution detects more than 500 diverse object categories and runs completely on-device in real-time as part of the AI Gallery photo search engine in the Samsung S21 series.
Worked on challenging video understanding problems for upcoming flagship smartphones.

Deep LearningComputer VisionReal-time ProcessingObject Detection

Senior Software Engineer

Jul 2018 – Mar 2020 · 1 yr 8 mos

As a member of the Visual Understanding team:
Designed and trained resource (memory and FLOPs) efficient deep learning solutions for human-object, human-human, and pet-object interaction detection in images.
Developed solution was commercialized for on-device AI Gallery Search in the flagship Samsung S21 smartphone series.
Trained deep neural networks in Caffe for real-time and on-device instance segmentation in images. Developed solution was first commercialized in Samsung S10 series.

Deep LearningInstance SegmentationReal-time ProcessingComputer Vision

Indian institute of technology, madras

2 roles

Teaching Assistant

Jan 2018 – May 2018 · 4 mos

Teaching Assistant for Object Oriented Algorithms Implementation & Analysis course at IIT Madras.

Teaching Assistant

Aug 2017 – Dec 2017 · 4 mos

Teaching Assistant for Principles of Machine Learning course at IIT Madras.

Adobe

Research Intern

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

Introduced the notion of fairness in recommendations in two-sided platforms, where users have two personas - consumers and creators.
Built recommendation algorithm which ensures fairness to content creators and gives representatively diverse and personalized recommendations to consumers.
Filed two US patents and published a research paper on the innovations.

Recommendation AlgorithmsFairness in RecommendationsData Science