cv
General Information
Full Name | Tatiana Gaintseva |
Title | AI Researcher |
Languages | English, Russian |
Academic Interests
- Computer vision, diffusion models, explainability, structure representation, knowledge extraction
Experience
-
Sep 2018 — presentFounding member, methodologist and lecturer , Deep Learning School
DLSchool is a non-profit organization that provides free online courses on AI fundamentails. It has over 4.000 learners every year. It's alumnus work in major institutions and companies, such as Stanford University, Moscow Institute of Physics and Technology and more.
I am one of the founding members of the school. Since the beginning, I lead a team that designes the methodology of the course and prepares lectures. I also give lectures on various AI topics, which have over 100.000 views in total.
-
Jul 2021 — Jun 2022Data & AI Scientist , Philips Innovation Labs RUS
I worked on developing novel AI-based tecnhiques for the software of the medical equipment that Philips procuses.
I was also a founder and a manager of scientific reading club, where people gather weekly to discuss recent advances in the field of AI for medicine.
-
Jan 2020 — Jun 2021AI Researcher , Huawei Rusian Research Institute
I was leading a group that conducted research in the area of AI-based face recognition. We managed to raise performance of an AI pipeline by 2.5%.
I was also managing a Huawei-MIPT colaboration on the task of developing domain adaptation techniques for AI-based face recognition. I reviewed the work of MIPT group and managed communications.
-
Nov 2018 — Jul 2021AI Researcher , GetRealPrice
GetRealPrice is a cutting-edge digital tech firm specializing in the creation of a B2B SaaS solution for tracking e-commerce pricing and executing product matching. By leveraging Big Data and AI-driven algorithms, the platform efficiently identifies comparable items across various online retailers, enabling seamless processing and analysis.
I was doing a research on AI techniques for the company business products. Models developed during this research resulted in substantial revenue for the company. I was also managing a group of researchers on one of the projects
-
Feb 2018 — Sep 2020AI Research Assistant , LAMBDA lab
I did research on applying generative models to high energy physics. Designed and implemented a novel pipeline for synthetic particle events generation. Had two scientific articles published, including one on ICRL workshop.
-
Apr 2017 — Aug 2017AI Research Intern , Yandex
Applied Reinforcement Learning to different metrics to seq2seq vocalization task. Combined different seq2seq vocalization models unsing ideas from Actor-Mimic algorithm. Showed that RL could increase quality of seq2seq models. This work was a part of bachelor’s diploma
Education
-
Nov. 2022 — presentPhD , Queen Mary University of London (QMUL)
DeepMind Studentship
Thesis title: Semantic Control in Denoising Diffusion Probabilistic Models
-
2017 — 2019MSc , Moscow Institute of Physics and Technology (MIPT)
Specialized in machine learning and data analysis.
GPA 7.1/10
Thesis title: Use of Domain Adaptation to expand the scope of Generative Models
-
2018 — 2020Master's level program , Yandex School of Data Analysis (YSDA)
Took courses on machine learning, different aspects of deep learning, applied statistics and big data.
-
2013 — 2017BSc , Moscow Institute of Physics and Technology (MIPT)
Specialized in mathematics and machine learning
GPA 8.4/10, magna cum laude
Thesis title: Multi-Objective Deep Reinforcement Learning in Seq2Seq Machine Translation
Research Publications
-
2020
Maxim Borisyak, Tatiana Gaintseva, and Andrey Ustyuzhanin, Adaptive divergence for rapid adversarial optimization. (2020). PeerJ Computer Science, 6, e274.
-
2020
Constantin Weisser, Daniel Craik, Tatiana Gaintseva, Artem Ryzhikov, Andrey Ustyuzhanin, Mike Williams, Autoencoders for Compression and Simulation in Particle Physics. ICLR FSAI 2020
Teaching
-
2022 — present
-
2019 — presentTeaching Assistant , Yandex School of Data Analysis
Leading seminars, preparing and reviewing students’ homeworks on deep learning class
-
2022Lecturer and methodologist , Open Machine Learning Course
This is an online free course on machine learning fundamentals. I was designing a methodology and preparing lectures for some topics.
-
2020 — 2022Lecturer and methodologist , Mathshub
I was a main methodologist and lecturer of computer vision fundamentals part of the course.
-
2020 — 2022Lecturer and methodologist , Pacticing Futures
During three years, I was designing multiple courses organized by Practicing Futures, and giving lectures on various deep learning topics
Competitions and Awards
-
Jun 2019MCS 2019 competition , organized by VisionLabs
3rd place
I was a part of a team that developed solution for competition on face recognition. Code on GitLab
-
Jun 2018MCS 2018 competition , organized by VisionLabs
2nd place
I was a part of a team that developed solution for competition on black-box adversarial attacks. Code on GitHub
-
Nov 2017Mood Map Project , Local Hack Day Hackathon, Moscow, Russia
winner
I was part of a team that developed a service that shows mood map of the city using Twitter and Yandex Maps APIs. Project on Devpost
-
Apr 2017
Public talks and posts
-
Sep 2022Structural representations in neural networks , article on habr.com
The article covers what structural representations is, why it is beneficial for neural networks and how to inject structural representations in them.
15k views
-
Nov 2021Inductive bias in neural networks , article on habr.com
The article tells what inductive bias is, what types of inductive biases there are in machine learning algorithms and in neural networks
13k views
-
Mar 2021Face recognition: how it works , article on RBC Trands
The article is about tech details of AI-based face recognition pipeline and how face recognition models are used in the world.
RBC is the large media in Russian language with 34 million monthly views
-
Jun 2019Machines Can See 2019 , conference talk
An oral talk about technical details of AI-based solution of MCS 2019 competition (3nd place). Competition page
over 2k views in total (online and offline)
-
Jul 2018Yandex ML Training , public talk
A public talk about technical details of AI-based solution of MCS 2018 competition (2nd place). Competition page
1.4k views
-
Jun 2018Machines Can See 2018 , conference talk
An oral talk about technical details of AI-based solution of MCS 2018 competition (2nd place). Competition page
over 2.5k views in total (online and offline)