About

Welcome to my blog! I’m Anton Dergunov, an avid practitioner of machine learning and AI, with a strong background in data science and software engineering based in London, United Kingdom. I’ve dedicated this space to share my insights, experiments, and discoveries in these exciting fields.

My Journey in Machine Learning

I’ve been fascinated by ML long before it became a trending topic. My interest in the field stems from several key motivations:

1. Data Insight: I love uncovering insights from data, making it a powerful tool for learning and discovery. The ability to visualize and explain data patterns is especially captivating to me, and I often use Jupyter Notebooks for their quick iteration capabilities.

2. Understanding the Brain: My curiosity about how the brain works also drives my interest in ML and AI.

3. Problem-Solving with AI: As a true geek, I find immense satisfaction in using AI tools to solve practical problems (such as helping to organize the information that I have or be a better writer).

Professional Background

My professional journey has been quite diverse and enriching. I earned my PhD with a focus on using knowledge representation techniques to encode expertise for fine-tuning software applications. This research led to multiple practical application which I integrated into several Intel software products. During my time at the university, I had the opportunity to teach AI courses and develop the course content.

I have worked with leading organizations such as Intel, Yandex, and Microsoft. At Intel, I focused on software development and research for high-performance and parallel optimization. At Yandex, I delved into ML projects like improving video ranking in their search engine. Currently, as a Senior Applied Scientist at Microsoft, I drive the integration of new ML ideas, from implementation and offline experimentation to online testing and model training in various products. Additionally, I have experience in building ML pipelines from the ground up to support these activities.

My expertise lies primarily in information retrieval and natural language processing. I have hands-on experience with various ML models, including linear models, decision trees, ensemble learning, and neural networks (including the transformer architecture). A particular area of interest for me is the interpretability of ML models, and I’ve developed internal tools at Microsoft for that.

My passion for ML extends beyond work hours, and I share many of my personal experiments on my GitHub repository.

Throughout my career, I have worked with a variety of platforms and technologies. Currently, I find Jupyter Notebooks and the Python ML stack (including tools like NumPy, Pandas, SciKit-Learn, PyTorch, and LightGBM, among others) particularly enjoyable, as they allow me to iterate quickly and effectively.

I have experience processing large volumes of data using technologies like PySpark, U-SQL, and custom-built MapReduce implementations. This includes tasks such as processing usage logs to prepare training and test data for ML models and extracting data from unstructured datasets using heuristics. In addition to these data-intensive applications, I also have substantial experience with compute-intensive parallelism and distributed processing using technologies like MPI, OpenMP, and others.

Research is a deep passion of mine. During my time at university, I attended numerous summer schools covering diverse topics such as information retrieval, semantic web, algorithms for massive data, formal software verification (in Moscow and Marseille), parallelism and concurrency, programming languages, etc. One of my notable contributions is an article on generalized algebraic data types which was reviewed by the creator of language, Simon Peyton Jones.

As a Unix enthusiast with experience in systems like FreeBSD, Solaris, and Linux, I greatly appreciate the command-line tools they offer. But for my daily activities, I choose to use macOS.

My approach to productivity and organization relies on Emacs Org mode for implementing the Getting Things Done (GTD) approach, which helps me manage projects, track tasks, and maintain a structured workflow. This tool has become an integral part of my life, enabling me to efficiently plan and execute my goals. Additionally, I maintain a comprehensive collection of personal notes in Markdown, which I’m transitioning to Obsidian to create a more organized “second brain” for accessing and managing my ideas and knowledge.

Personal Interests

Outside of my professional pursuits, I’m deeply committed to personal growth and continuous learning. I try to make the most of every free moment, including my daily commute, to read and stay informed. Whether it’s diving into books, articles, or research papers, I strive to turn even brief periods into opportunities for learning and enrichment.

I have a passion for languages. While I am fluent in English and Spanish, I have studied French, Italian, and German in the past but haven’t practiced them recently. I plan to revisit these languages to refresh and improve my skills.

Whenever I get the chance, I love traveling and immersing myself in nature. Some of my most cherished experiences include exploring Kyoto in Japan, taking a scenic train journey in Sri Lanka, visiting the Kon-Tiki Museum in Oslo, and marveling at the Alishan National Scenic Area in Taiwan. The highlight of my hiking adventures was the solo Tour du Mont Blanc, though the most meaningful hikes were those with my parents during my childhood. My understanding of photography and videography is more technical than artistic at the moment, but I’m casually developing my creative skills as a hobby.

When it comes to relaxation, I enjoy savoring good coffee and exploring different teas. From rich, aromatic coffees to authentic Chinese, Taiwanese, and Japanese teas, I enjoy discovering and appreciating their unique flavors. To stay healthy, I dedicate some reasonable time to gym workouts, cycling, and yoga.

Feel free to connect with me on LinkedIn and share your thoughts on my blog posts. Happy reading!