I’m Konstantin Nikiforov — MD, Molecular Geneticist, and Machine Learning Engineer.

I build end-to-end ML systems: data acquisition → feature engineering → modeling → evaluation → APIs → user-facing applications → reproducible pipelines.

Focus: applied ML, information retrieval & semantic search, recommendation systems, ML platforms, reproducible experimentation, and model evaluation.

Interests: R&D, MLOps, biomedical ML, NLP, CV, recommender systems, interpretable ML, time series, and modern deep learning.

ML / Data Science: Python · SQL · Pandas · NumPy · scikit-learn · CatBoost · XGBoost · LightGBM · Optuna

Retrieval / NLP: Sentence Transformers · SBERT · embeddings · semantic search · hybrid lexical+dense retrieval · Qdrant · Hugging Face

Deep Learning: PyTorch · Hugging Face Transformers

Recommender Systems: ALS / implicit feedback · content-based retrieval · hybrid recommenders · learning-to-rank

MLOps / Backend: FastAPI · Docker / Docker Compose · Airflow · MLflow · experiment tracking · validation pipelines

Data / Storage: PostgreSQL · MySQL · PySpark · Parquet · vector databases

Applications / Visualization: Streamlit · Plotly · Dash · Matplotlib · SHAP · LIME

Time Series: Prophet · TBATS · ETNA · AutoTS · rolling / holdout backtesting

Domains: Tabular ML · Information Retrieval · Recommenders · NLP · Time Series · BioML · Computer Vision

Exploring: RAG · LangChain / LangGraph · GNNs · scientific embeddings & cross-encoders · ESM / protein language models · self-supervised learning · Diffusion · Ray / Dask · Kafka · Kubernetes · observability (Prometheus / Grafana / OpenTelemetry)

Rust · C · C++ · Java

Research intelligence platform for discovering, organizing, and analyzing Machine Learning and AI research.

Built an end-to-end multi-source data pipeline integrating arXiv, OpenAlex, Crossref, Semantic Scholar, and ACL Anthology into a reconciled corpus of ~61K canonical papers.

Implemented:

• metadata ingestion, normalization, reconciliation, and provenance tracking;
• lexical, dense, and hybrid semantic retrieval;
• Sentence Transformer embeddings and Qdrant vector search experiments;
• topic clustering and similar-paper retrieval;
• citation/reference and paper–artifact graph analytics;
• PostgreSQL-backed serving layer;
• FastAPI backend and Streamlit research workspace;
• automated data-quality, regression, and contract validation.

repo

Hybrid recommendation system combining CatBoost + ALS + SBERT, with FastAPI, Streamlit, Docker, and Qdrant.

repo

End-to-end conversion prediction pipeline using a calibrated CatBoost / XGBoost / LightGBM ensemble, FastAPI, Streamlit, Airflow DAGs, and Docker Compose.

repo

Monitoring pipeline with Evidently, SHAP, PSI, and Jensen–Shannon divergence, including configurable alert policies.

repo

Forecasting experiments with ARIMA, TBATS, Prophet, Darts, time-aware validation, and Optuna-tuned baselines.

repo

Bioinformatics project exploring RNA-seq dimensionality reduction and representation learning for survival analysis.

repo

English — B2 · French — B2

Email: konnik1000@gmail.com Telegram: @Konnik1988 GitHub: https://github.com/KonNik88

Machine Learning Engineer with a molecular genetics / biomedical background, building end-to-end ML systems across data engineering, retrieval, modeling, evaluation, APIs, and deployment.

Currently developing ML Research Radar — a multi-source research intelligence platform with semantic search, hybrid retrieval, graph analytics, PostgreSQL, FastAPI, Streamlit, and reproducible data/ML pipelines.