← В ленту
Senior
Регистрация: 02.01.2020

Роман Куцев

Специализация: Crowd Solution Architect
Размечаю данные на заказ, помогаю с настройкой краудсорсинговых проектов в Яндекс.Толоке и Amazon Mechanical Turk. Более 8 лет в области DS и более 6 лет в области сбора данных и аннотирования. Руководил штатом разметчиков в Prisma. Являюсь сертифицированным партнером Яндекс.Толоки. Сооснователь компании TrainingData.Solutions. Вместе с соучредителями мы создали компанию с более чем 1000 аннотаторами и доходом в несколько миллионов долларов. Реализовал более 200 проектов по разметке данных для обучения нейронных сетей. Преподаю краудсорсинг в ШАДе, ВШЭ, МФТИ.
Размечаю данные на заказ, помогаю с настройкой краудсорсинговых проектов в Яндекс.Толоке и Amazon Mechanical Turk. Более 8 лет в области DS и более 6 лет в области сбора данных и аннотирования. Руководил штатом разметчиков в Prisma. Являюсь сертифицированным партнером Яндекс.Толоки. Сооснователь компании TrainingData.Solutions. Вместе с соучредителями мы создали компанию с более чем 1000 аннотаторами и доходом в несколько миллионов долларов. Реализовал более 200 проектов по разметке данных для обучения нейронных сетей. Преподаю краудсорсинг в ШАДе, ВШЭ, МФТИ.

Портфолио

Prisma Labs, Inc.

R&D Engineer. Recruiting employees on freelance platforms and supervising them creating large labeled image datasets. Creating tasks for labelling data on Amazon MTurk and Yandex Toloka. Programming utilities for faster and more suitable work with datasets. Results: • Completed 8 projects and created 21 datasets. • Marked over 2 000 000 images. • Created important datasets for training neural networks for segmentation and face detection tasks. These neural networks are used in the Lensa App.Recruiting employees on freelance platforms and supervising them creating large labeled image datasets. Creating tasks for labelling data on Amazon MTurk and Yandex Toloka. Programming utilities for faster and more suitable work with datasets. Results: • Completed 8 projects and created 21 datasets. • Marked over 2 000 000 images. • Created important datasets for training neural networks for segmentation and face detection tasks. These neural networks are used in the Lensa App. Data Scientist Intern. Provided verification datasets, created a technical specification for datasets markup.

Neatsy, Inc.

I was responsible for creating datasets for training neural networks and was responsible for all data collection in the company. Achievements: - Gathered a focus group of 500 people trying on shoes. Based on the collected data, our recommendation system predicted the correct size for 300+ shoe models. - Set up processes for storing and working with data in the company. - Created a dataset of 3D foot scans. - Collected more than 50,000 foot images on crowd sites.

Impressive bot

Show project. Chatbot describing the first impression of a person.

Скиллы

Machine learning
Data collection and annotation
Business growth
Cryptocurrencies
Creation of services and products
Python
Data Science
JavaScript
HTML/CSS
Docker
OpenCV
PostgreSQL

Опыт работы

Lecturer
с 09.2020 - По настоящий момент |Yandex School of Data Analysis
.
Lecturer at Yandex School of Data Analysis. Course "Collecting and Marking Data for Machine Learning".
Сrowd Solution Architect
05.2019 - 12.2022 |Neatsy, Inc.
.
I was responsible for creating datasets for training neural networks and was responsible for all data collection in the company. Achievements: - Gathered a focus group of 500 people trying on shoes. Based on the collected data, our recommendation system predicted the correct size for 300+ shoe models. - Set up processes for storing and working with data in the company. - Created a dataset of 3D foot scans. - Collected more than 50,000 foot images on crowd sites.
Co-Founder & CTO
с 10.2018 - По настоящий момент |TrainingData
.
Training Data on Demand. Having high quality training data is necessary for training a neural network. Our team takes all responsibility for creating a dataset for your task.
R&D Engineer, Data Scientist Intern
12.2017 - 05.2019 |Prisma Labs, Inc.
.
R&D Engineer. Recruiting employees on freelance platforms and supervising them creating large labeled image datasets. Creating tasks for labelling data on Amazon MTurk and Yandex Toloka. Programming utilities for faster and more suitable work with datasets. Results: • Completed 8 projects and created 21 datasets. • Marked over 2 000 000 images. • Created important datasets for training neural networks for segmentation and face detection tasks. These neural networks are used in the Lensa App.Recruiting employees on freelance platforms and supervising them creating large labeled image datasets. Creating tasks for labelling data on Amazon MTurk and Yandex Toloka. Programming utilities for faster and more suitable work with datasets. Results: • Completed 8 projects and created 21 datasets. • Marked over 2 000 000 images. • Created important datasets for training neural networks for segmentation and face detection tasks. These neural networks are used in the Lensa App. Data Scientist Intern. Provided verification datasets, created a technical specification for datasets markup.
Аnalyst Intern
07.2016 - 08.2016 |Tinkoff Bank
SQL
I analyzed incoming customer calls and identified the main problems when using the mobile application and the bank's website. Achievements: - described 3 problems in business processes, the solution of which made it possible to reduce the load on the call center by 7%; - figured out the data warehouse, made SQL queries, to get up-to-date statistics on the types of calls to the call center and its workload.

Образование

Operations Research
2018 - 2020
Lomonosov Moscow State University (MSU)
Quantitative Financial Analytics
2015 - 2016
CMF
Applied Mathematics and Computer Science
2013 - 2018
Lomonosov Moscow State University (MSU)

Языки

РусскийРоднойАнглийскийВыше среднего