Skip to content
View kirilklein's full-sized avatar
:octocat:
:octocat:
11 followers · 24 following

Achievements

Achievement: Pair Extraordinairex3Achievement: Pull Sharkx3Achievement: QuickdrawAchievement: YOLO

Achievements

Achievement: Pair Extraordinairex3Achievement: Pull Sharkx3Achievement: QuickdrawAchievement: YOLO

Block or report kirilklein

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. Markdown is supported. 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
kirilklein/README.md

Kiril Klein

Machine Learning Engineer · Causal Inference & Clinical NLP · PhD

LinkedIn Google Scholar Email

I build ML systems for clinical data: transformers pre-trained on longitudinal electronic health records, clinical NLP pipelines, and causal-inference methods for estimating treatment effects from observational data. PhD in Machine Learning and Causal Inference, University of Copenhagen.

Work

Aiomic (Copenhagen) — building the product ML stack from scratch: transformers combined with classical NLP/ML in a single pipeline, from data preparation and training through validation to Docker-based services and CI/CD.

PhD, University of Copenhagen — Scalable Causal Inference on Electronic Health Records Using Transformers. Pre-training transformers on structured longitudinal EHR and using them for outcome prediction and treatment-effect estimation.

Currently — maintaining CausalEstimate and contributing treatment-effect fixes to statsmodels.

Selected publications

CORE-BEHRT: A Carefully Optimized and Rigorously Evaluated BEHRT
PMLR vol. 252 (MLHC), 2024. Joint first author.

MEDS — An Emerging Data Standard and Ecosystem for Health AI Research
NEJM AI, 2026. Co-author. (website · GitHub org)

BONSAI: A framework for processing and analysing Electronic Health Records (EHR) data using transformer-based models
Journal of Open Source Software, 2025. Co-author.

Full list on Google Scholar.

Python · PyTorch / Lightning · Hydra · scikit-learn · Dask · Docker · Azure · GitHub Actions

Some professional work lives under my work account, @kvk-cmd.

Pinned Loading

  1. FGA-DIKU/BONSAI FGA-DIKU/BONSAI Public

    A BERT-based framework for processing and analyzing Electronic Health Records (EHR) data. It provides an end-to-end pipeline for data preprocessing, model training, and clinical outcome prediction.

    Python 11 6

  2. causal-sandbox causal-sandbox Public

    Learn causal inference by changing the world. Guided lessons and a sandbox with confounders, mediators, colliders, and six estimators, live in the browser

    JavaScript

  3. mikkelfo/CORE-BEHRT mikkelfo/CORE-BEHRT Public

    Jupyter Notebook 19 5

  4. CausalEstimate CausalEstimate Public

    Lightweight causal inference from precomputed propensity scores and outcome predictions: IPW, AIPW, TMLE, matching, bootstrap CIs. Pandas-native.

    Python 2 2

  5. bonsai-causal bonsai-causal Public

    BONSAI Causal: transformer (ModernBERT) pipeline for EHR data in MEDS format, extended for causal inference — cohort matching, joint exposure/outcome finetuning, calibration, IPW/AIPW/TMLE effect e…

    Python 2 2

  6. mergeprint mergeprint Public

    A privacy-first, interactive one-page dashboard of your year on GitHub

    HTML