Пожалуйста, войдите, чтобы скачать резюме.
Experience level:
More than 5 years
Education level:
Master
Languages:
Russian: Native
English: C1
Serbian: B1
I’m a research-oriented machine learning engineer working at the intersection of modeling, data infrastructure, and algorithms.
I build and improve ML data pipelines, evaluation workflows, and engineering systems for technically demanding problems. At Yandex.Translate, I worked on multilingual machine translation data pipelines over web-crawled corpora, alignment heuristics for low-resource language pairs, and automated refresh and evaluation workflows at petabyte scale across 100+ language pairs. At Acadé Studio, I contribute algorithmic reasoning and code-generation data for LLM training, combining structured solution traces, C++ implementations, tests, and multi-stage review.
My core strengths are Python, C++, algorithms, and evaluation-heavy work on noisy data. I’m especially interested in NLP, LLM data and evaluation, ranking and recommendation-adjacent ML, reinforcement learning, and research-oriented engineering where careful experimentation matters.
I also bring scientific rigor and mathematical depth from an astrophysics background, including experience with simulation, numerical methods, and technically demanding research problems.
I’m most interested in ML Engineer / Research Engineer roles where modeling, infrastructure, and rigorous experimentation all matter.
• Produced competitive-programming task packages for reasoning and code-generation LLM data across 10+ online judges, combining statement normalization/translation, structured solution traces, incremental C++ implementations, and custom tests.
• Built and reviewed a corpus of 300+ task packages and 400+ review records, documenting edge cases, correctness considerations, and failure modes to improve dataset reliability.
• Participated in multi-stage review, challenge, and dispute-resolution workflows, evaluating solution correctness, clarity, and failure modes for quality control.
• Worked on independent programming projects spanning automation, small-scale database and server-side tasks, exploratory implementations, and data-processing work.
• Provided private tutoring in mathematics, machine learning, physics, and astrophysics, including olympiad-oriented problem solving and mentoring.
• Built and maintained backend pipelines for multilingual machine translation data used by Yandex.Translate and related model-development workflows.
• Processed web-crawled corpora from Yandex infrastructure and Common Crawl to extract and align bilingual sentence pairs.
• Reworked document-alignment heuristics using cross-link structures to improve data quality and recall for low-resource language pairs.
• Ran petabyte-scale workflows with YQL, YT, and Nirvana across 100+ language pairs, automating regular data refreshes and experimental evaluation.
• Designed advanced astronomy and physics curricula and original problem sets for Olympiad-level students.
• Mentored national and international prize winners and developed technical teaching materials and visualizations in Python and LaTeX.
• Improved Python APIs for interaction with core C++ infrastructure on YTsaurus, Yandex’s distributed data-processing platform.
• Worked on production backend tasks alongside infrastructure engineers, gaining early experience with large-scale distributed systems.
Python, C++, SQL, Linux, Git, Machine Learning, NLP, Machine Translation, LLM Evaluation, Model Evaluation, Data Pipelines, Data Engineering, Large-Scale Data Processing, Algorithms, PyTorch, Backend Development, Distributed Systems, Statistical Modeling, Optimization, Recommender Systems, Reinforcement Learning