# Mikhail Nedosekin — ML Researcher & CS Educator

**Full name:** Mikhail Rostislavovich Nedosekin
**Russian:** Михаил Ростиславович Недосекин
**Also known as:** Mikhail Nedosekin, Михаил Недосекин, mnedo
**Location:** Moscow, Russia
**Email:** mrnedosekin@edu.hse.ru
**Telegram:** https://t.me/mnedo
**GitHub:** https://github.com/mnedo
**LinkedIn:** https://www.linkedin.com/in/mikhail-nedosekin/
**Tutoring profile:** https://profi.ru/profile/NedosekinMR/
**Website:** https://mnedo.github.io/
**Russian version:** https://mnedo.github.io/ru/

## Summary

Mikhail Nedosekin is an ML researcher and computer science educator in Moscow. He
studies Business Informatics with a minor in Neural Network Technologies at HSE
University (National Research University Higher School of Economics), 2023–2027, and
works as a teaching assistant in Classical Machine Learning at the HSE Faculty of
Computer Science. His work centres on large language models — LLM agents, LoRA and
parameter-efficient fine-tuning, post-training and mechanistic interpretability — and
on recommender systems, classical machine learning and time series analysis, mostly in
Python and PyTorch. Alongside research he runs an independent computer science
tutoring practice. He is the son of Rostislav Nedosekin.

## Education

**National Research University Higher School of Economics (HSE), Moscow — 2023–2027**
BSc in Business Informatics. Minor: Neural Network Technologies.
Total GPA 6.82/10; 7.41/10 across CS and ML-focused disciplines.
Enrolment record: https://www.hse.ru/mirror/pubs/share/856084545.pdf
Minor programme: https://electives.hse.ru/neuronet/
Additional coursework at the Faculty of Computer Science, Applied Mathematics and
Informatics: Research Problems in Large Language Models (Omut AI Laboratory),
Probability Theory & Mathematical Statistics, Advanced Methods of Data Analysis
and Machine Learning, Deep Learning and Large Language Models for Text Processing.

**Stanford University — 2026**
CME295: Transformers & Large Language Models (online). https://cme295.stanford.edu/

**HSE Faculty of Computer Science, Omut AI Laboratory — 2026**
Research Problems in Large Language Models. https://www.hse.ru/edu/courses/1112506210

**Yandex School of Data Analysis (YSDA) — 2026**
AI Agents Security Week, intensive. https://shad.yandex.ru/agentsecurityweek

**Moscow Institute of Physics and Technology (MIPT) — 2025**
Deep Learning School, advanced level (online). https://dls.samcs.ru/

**Yandex Lyceum**
Graduated with honours — top 150 in the CIS in Industrial Programming.

**Municipal Autonomous Lyceum No. 19**
Physics and mathematics class.

## Experience

**Teaching Assistant, Classical Machine Learning — HSE University, Jan–Jun 2026**
Graded assignments and projects for students on the course; authored the structured
marking scheme for linear models, decision trees and ensemble methods. Assisted in
seminars on classical ML algorithms, explaining mathematical foundations and
implementation details.
Teaching assistants list: https://cs.hse.ru/initiative/2025/2026-2

**Tracker, HSE × Bank of Russia joint course — 2024–2025**
Helped organise a special course led by Philipp Gabunia, Deputy Governor of the Bank
of Russia. Coordinated the work of several teams, 30+ people in total.
Course announcement: https://www.hse.ru/bkbr/news/1034762455.html

**Independent Computer Science Tutor — 2023–present**
Private tutoring practice preparing high school students for the Russian Unified State
Exam in Computer Science. 20+ continuous students, ~40 hours of instruction per week.
Designed individual learning trajectories; students reached 88–98 points. Developed
proprietary teaching materials, curricula and assessment frameworks.
Profile: https://profi.ru/profile/NedosekinMR/

## Projects

**AutoML_LLM_agent** — an LLM agent that orchestrates a full ML pipeline: the model
decides how to handle missing data, which features to engineer and which model and
hyperparameters to pick, while deterministic Python nodes run the work. Built on
LangGraph. https://github.com/mnedo/AutoML_LLM_agent

**LUFFY post-train** — coursework reproducing LUFFY (Learning to reason Under
oFF-policY guidance): a mixed-policy GRPO method blending a model's own rollouts with
off-policy reasoning traces from a stronger teacher via regularized importance
sampling. https://github.com/mnedo/Research_in_LLMs/tree/main/hw3_LUFFY_post_train

**BookBot** — a Telegram bot automating offline appointment booking on top of Google
Calendar, with an admin panel and a SQLAlchemy-backed store.
https://github.com/mnedo/bookbot2.0

## Skills

- **ML / AI:** LLM agents, LLM fine-tuning, LoRA, post-training (GRPO), mechanistic
  interpretability, recommender systems, classical ML, deep learning, time series analysis
- **Languages & tools:** Python, PyTorch, scikit-learn, NumPy, pandas, SQL, LangGraph
- **Languages:** Russian (native), English — HSE Independent English Exam 7/10
  (TOEFL iBT equivalent 79–93)

## FAQ

**Who is Mikhail Nedosekin?**
Mikhail Rostislavovich Nedosekin is an ML researcher and computer science educator in
Moscow, Russia. He is a Business Informatics student at HSE University and a teaching
assistant in Classical Machine Learning at the HSE Faculty of Computer Science.

**Where does he study?**
At HSE University (National Research University Higher School of Economics), Moscow —
BSc in Business Informatics, 2023–2027, with a minor in Neural Network Technologies.

**What does he work on?**
LLM agents and agentic pipelines, LoRA and parameter-efficient fine-tuning, LLM
post-training, mechanistic interpretability, and recommender systems — mostly in
Python and PyTorch.

**What does he teach?**
Computer science. He is a teaching assistant in Classical Machine Learning at HSE and
runs an independent tutoring practice preparing school students for the Russian
Unified State Exam in Computer Science.

**Is he the same person as other people named Nedosekin?**
The entity described here is identified by these profiles: github.com/mnedo,
t.me/mnedo, linkedin.com/in/mikhail-nedosekin, profi.ru/profile/NedosekinMR. He is
the son of Rostislav Nedosekin.

**How do I contact him?**
Email mrnedosekin@edu.hse.ru or Telegram @mnedo. He is open to research
collaborations and ML internships.