Rutgers University

Machine Perception and AR/VR at Facebook

I lead a multinational organization of computer vision and machine learning scientists, engineers and managers across the US and Europe. We develop machine perception technologies for human understanding that power key product experiences in Augmented Reality (AR) and Virtual Reality (VR).

We work full-stack across applied research and development of algorithms, performance optimizations for target hardware, and production deployments into target devices and apps. We build on the state of the art methods in computer vision and machine learning including (semi-)supervised deep learning, generative adversarial learning, active learning, depth sensing, 3D reconstruction, and rendering. We also invest in developing solid technological foundations with well-structured software frameworks, systems for data storage and privacy handling, and workflows for model development and responsible AI.

Our teams are growing. We are hiring scientists, engineers and managers with expertise in computer vision and machine learning.

Computer Vision and Machine Learning for AR/VR

Led a 30+ person organization of scientists, engineers and managers working on machine learning and recommendation systems. We developed break-through personalization technologies and new products powering advertising across Facebook, Instagram and all other surfaces.

Our system makes personalized connections between over two billion people, millions of businesses and billions of products globally. We achieve this through research and development of deep learning models for retrieval and ranking using massive-scale multimodal data (text, images), user intent understanding, counterfactual methods, and other machine learning techniques.

Filed 5 patents in recommendation systems, deep learning, and new advertising products.

Machine Learning for Ads

Led a team of modeling scientists and engineers personalizing LinkedIn news feed - making it engaging for over 400 million LinkedIn members while driving hundreds of millions in annual revenue. By developing machine learning methods for personalized feed ranking, we increased user engagement by 40% and helped grow annual revenue by over 2x.

(Co-)Inventor of 18 US patents spanning machine learning for content recommendations, educational decision making, and personalization of user interfaces in social networks.

LinkedIn

Head of Machine Learning for News Feed Personalization

Led a team of machine learning scientists and engineers building technologies for optimal allocation of online marketing budgets at X million pricing decisions per second. We developed innovative machine learning solutions for ad conversion prediction, pricing, fraud detection, and data driven systems for managing ad campaigns. We drove double-digit increases in performance of Quantcast's Advertising product and company revenue.

(Co-)Inventor of 4 US patents on machine learning for personalization and optimal delivery of online advertising campaigns.

Quantcast

Machine Learning Scientist and Manager, Ads

Researched and developed causal discovery and machine learning algorithms for phenotype prediction and identification of underlying causal factors using 60+ genomics datasets (microarrays, micro RNA arrays and genome-wide SNP arrays).

Published a book chapter, and coauthored 7 research publications in leading journals including Journal of Machine Learning Research (JMLR), Cell Host & Microbe, Genomics.

NYU Langone Health

Machine Learning Scientist, Genomics

Developed machine learning approaches for automatic detection and segmentation of ischemic patterns from cardiac perfusion magnetic resonance imaging (MRI) time series. Clustering in spatial and temporal domains. Implemented graph and kernel-based clustering algorithms. Conducted a broad benchmarking evaluation of unsupervised learning methods for segmentation of ventricular blood pools and the myocardium.

Siemens

Research Intern, Machine Learning, Medical Imaging and Visualization

Developed Bayesian Networks for diagnosis and prediction of faults in semiconductor manufacturing process.

Intel Corporation

Reserarch Intern, Machine Learning, Analysis & Control Technology

Developed Vicinal Support Vector Machines with transformation-invariant kernel functions for hand written character recognition.

NEC Corporation

Research Intern, Machine Learning

Developed novel methods for clustering data by gradient-based combinatorial optimization of objective functions involving higher-order statistical moments. Designed and implemented highly-scalable clustering algorithms for applications on financial time series and text datasets. Coauthored a National Science Foundation research grant proposal.

Nikita Lytkin