Scholar
David Dale
Google Scholar ID: 4GB_6AcAAAAJ
Meta AI
Citations & Impact
Citations
976
H-index
14
i10-index
16
Publications
20
Co-authors
21
list available
Publications
Resume (English only)
Academic Achievements
- Built the first neural machine translation system for the Erzya language (published paper, Russian blog post, open-source code)
- Created Dialogic, a Python framework for building chatbots
- Developed compress-fasttext, a package for compressing FastText models
- Published Russian-language neural models (e.g., tiny BERT, Russian T5) on Hugging Face
- Had a paper on text style transfer accepted at EMNLP
Research Experience
- Works at Meta’s FAIR (Fundamental AI Research) on NLP projects including 'No Language Left Behind' and 'Seamless: Multilingual Expressive and Streaming Speech Translation'
- Previously worked as a research developer in the NLP Lab at Skoltech University, focusing on text style transfer
- Contributed to the development of Yandex’s Alice voice assistant—improved scenarios for Yandex.Station, enhanced intent classifier quality, and fixed (and introduced) bugs
- Served as an analyst at Yandex Data Factory, helping industrial clients apply AI to optimize production (e.g., accelerated pipe hardening at Chelpipe without quality loss)
- Worked in Retail Risk Management at Alfa Bank—developed credit loss forecasting models, improved credit card issuance strategies, and optimized debtor calling algorithms
- Teaches 'Probability and Statistics for Data Scientists' at Y-DATA (Tel Aviv School of Data Analysis)
- Co-teaches 'Introduction to Data Science for Product Managers' with MIPT and MathsHub
- Delivers ad-hoc lectures on problem formulation in data science, NLP, mathematical modeling fundamentals, and related topics
Background
- IT specialist primarily working on natural language processing (NLP)
- Originally from Russia, currently living in Paris
- Opposes Russian military aggression and supports Ukraine's independence and territorial integrity
- Supports anti-war and liberation movements within the Russian Federation through donations and volunteer work
- Has a full-time job but open to interesting and ethical side projects
Miscellany
- Runs the English Substack newsletter 'Möbius duct tape'
- Writes technical blog posts on Habr (Russian) and Medium (English)
- Maintains several Telegram channels and chats: nlp_jobs (NLP jobs), 'Изолента Мёбиуса' (programming/NLP in Russian), 'Матчасть' (applied math in Russian), and 'Botcamp' (dialogue tech chat in Russian)
- Shares technical talks on YouTube, including building multiskill voice assistants, NER in Alice (in Russian), and feature engineering basics (in Russian)
- Published the 2016 'Матчасть' lecture series (30 hours of introductory applied math and data analysis)
- Contact: Telegram, VK, Facebook, Twitter, email daledavidd@gmail.com
