Skip to content
View IlyaShaposhnikov's full-sized avatar
  • Russia, Saint Petersburg

Block or report IlyaShaposhnikov

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
IlyaShaposhnikov/README.md

🇷🇺 Russian Version / На русском

👋 Hi, I'm Ilya Shaposhnikov

ML/NLP/LLM Engineer building production-grade machine learning systems and intelligent language applications. I combine a linguistics background with strong software engineering to bridge the gap between research notebooks and real-world deployment.

I focus on building robust, production-ready ML pipelines — from text preprocessing and model training to rigorous evaluation, explainability (SHAP), containerization, CI/CD, and deployment. Below are projects I architected and implemented: a GitHub-issue triage system comparing fine-tuned transformers, embeddings and TF-IDF and using dense retrieval for duplicate search, a comparative topic-modeling study (LDA/NMF/LSA) on 20K arXiv abstracts, an end-to-end sentiment pipeline comparing TF-IDF and sentence embeddings with logistic regression and a PyTorch MLP, a local-LLM analytics bot, and a multilingual extractive summarizer. Four of them run as live demos.

Why NLP & ML?

My path into ML wasn't traditional — and that's my edge. With a background in linguistics and Eastern languages, I spent 12+ years working with language data at an international corporation: managing NMT workflows, CAT systems, terminology databases, and translation memory. I saw firsthand how language data flows, where linguistic bottlenecks hide, and how automation saves thousands of hours.

This deep understanding of how language works — from morphology to semantics — naturally pulled me into Machine Learning. Today, I combine this linguistic intuition with modern ML engineering to build systems that truly understand human language.

🚀 Key Projects

Category Project Key Technologies Core Concept & Challenges
NLP & Semantic Search 🧭 TriageLens Transformers (DistilBERT), Sentence-Transformers, Chroma, BM25, scikit-learn, Streamlit Issue-type classification and duplicate search on 38K live GitHub issues (uv, pandas): dense retrieval (bge-small) finds the original in the top 10 for 52–58% of duplicates vs 27–33% for BM25; fine-tuned DistilBERT vs TF-IDF vs zero-shot with leakage removal and a time-based split. Live demo with weekly index refresh, 140+ tests.
NLP Research 🔎 TopicLens scikit-learn, NLTK, Plotly, Streamlit, pytest Comparative study of LDA, NMF and LSA on 20K arXiv abstracts (2018–2026): coherence, topic diversity and agreement with real categories, Hungarian topic matching across models, 8-year topic dynamics. Live demo with monthly auto-refresh, 280+ tests.
NLP Pipeline ✈️ Airline Sentiment Analysis scikit-learn, PyTorch, Sentence-Transformers, FastAPI, Streamlit, SHAP TF-IDF vs sentence embeddings × logistic regression vs PyTorch MLP over 5 seeds: embeddings add +2–3 pp macro-F1 at ~40× latency. Confidence-weighted training, SHAP explanations, REST API, multi-model dashboard. Live demo, 290+ tests.
LLM Application 🤖 Video Analytics Bot Aiogram, Ollama (LLM), PostgreSQL, asyncpg NLP-powered bot: transforms natural language queries into SQL analytics using a local LLM (Mistral 7B) with prompt engineering and few-shot examples.
NLP Application 🌍 Polyglot Extractive Summarizer sumy (TextRank), NLTK, Streamlit, wikipedia-api Multilingual (5 languages) extractive summarizer for Wikipedia articles and local files with flexible length control, Streamlit web UI with live demo, and CLI.
Backend & API 💰 Wallet REST API FastAPI, PostgreSQL, Docker, async Asynchronous API for financial operations with guaranteed data consistency under concurrent requests (transactions, row-level locks). Backend groundwork for serving ML models under concurrent load.

▶️ Live Demos: 🧭 TriageLens · 🔎 TopicLens · ✈️ Airline Sentiment Analysis · 🌍 Polyglot Extractive Summarizer

📚 Full Project List — all projects with detailed descriptions

🛠 Technology Stack

Category Technologies
ML, NLP & LLMs scikit-learn pandas NumPy Gensim NLTK PyTorch Sentence-Transformers Transformers Hugging Face Hub Ollama SHAP SciPy Matplotlib Seaborn Plotly Streamlit
Backend & MLOps Python FastAPI Django Django REST Framework Flask Aiogram SQLAlchemy Alembic Pydantic
Databases PostgreSQL asyncpg MySQL SQLite Chroma
DevOps, Testing & Tools Docker Docker Compose Git GitHub Actions Nginx Gunicorn Bash Pytest
API & Security REST API JWT Postman GraphQL

🗣️ [Natural] Languages

  • 🇷🇺 Russian: Native
  • 🇬🇧 English: C2 (Proficient)
  • 🇪🇸 Spanish: C1 (Advanced)
  • 🇫🇷 French: B2 (Upper-Intermediate)
  • 🇩🇪 German: B1 (Intermediate)
  • 🇸🇦 Arabic: B1 (Intermediate)

Multilingual proficiency gives me a practical edge in NLP — I understand morphology, syntax, and semantics across language families firsthand.

🎯 Open to Opportunities

I'm looking for an ML/NLP/LLM Engineer role in a product-driven team where I can apply my unique blend of linguistic intuition, ML expertise, and production engineering to build intelligent systems that solve real-world problems.

📍 Location: Saint Petersburg, Russia (open to on-site, remote, and hybrid work)

📧 Contact: ilia.a.shaposhnikov@gmail.com

📱 Telegram: @iliashaposhnikov

💼 LinkedIn: @iliashaposhnikov

🇷🇺 Russian Version / На русском

Pinned Loading

  1. wallet-api wallet-api Public

    A FastAPI-based REST API for managing user wallet balances with secure concurrent transaction handling using SELECT FOR UPDATE to ensure data integrity, built with asynchronous PostgreSQL and Docker

    Python

  2. video_analytics_bot video_analytics_bot Public

    A Telegram bot that uses a local Ollama LLM to process natural language queries and retrieve video analytics from a PostgreSQL database

    Python

  3. airline-sentiment-analysis airline-sentiment-analysis Public

    End-to-end airline tweet sentiment analysis: TF-IDF vs. sentence embeddings with confidence-weighted Logistic Regression or a PyTorch MLP, benchmarked over 5 seeds; SHAP/weight explanations, FastAP…

    Python

  4. polyglot-extractive-summarizer polyglot-extractive-summarizer Public

    A multilingual extractive text summarizer using the TextRank algorithm via sumy and NLTK for Wikipedia articles and local .txt files, featuring adjustable summary length, original order preservatio…

    Python

  5. topic-lens topic-lens Public

    A comparative topic modeling toolkit that fits LDA, NMF and LSA on the same corpus of arXiv abstracts and measures them against each other with NPMI coherence, topic diversity and agreement with re…

    Python

  6. triage-lens triage-lens Public

    Triage for GitHub issues: predicts the issue type and finds earlier duplicates. Built on live uv and pandas issues, comparing TF-IDF, embeddings, fine-tuned DistilBERT and zero-shot NLI for classif…

    Python