Data Scientist
PhD · Learning Analytics
Tallinn, Estonia
Applied AI across
education, genomics
& health.
I build models people can trust and interfaces they actually use — from real-time classroom collaboration analytics to microbiome sequencing pipelines.
Now · Metrosert AS
Senior Data Scientist
National registry data access, medical health records analysis, bioinformatics pipelines and research proposals with healthcare partners.
Explore by domain
Projects, papers & posts
Project
Genomics & health
Genome sequencing data processing pipelines
Developed upstream and downstream pipelines to process amplicon sequencing data
Paper
Learning analytics
A quarter century of advances in collaboration sensing: a comprehensive review of high-frequency metrics and their connection to collaborative constructs from 2000 to 2025
International Journal of Computer-Supported Collaborative Learning
Post
Genomics & health
Is Your Health App a Medical Device? A Simple Guide to Software under EU MDR
A software application planned for healthcare domain in EU (European Union) countries comes under the same EU regulation sets out for medicinal physical products. However…
Project
Bayesian & maths
Bayesian modeling of students' mathematics skills
Processed and analyzed students' interaction data from a learning platform to model their mathematics skill mastery levels for 9t…
Paper
Data & visualization
Gut microbiome biomarkers for colorectal cancer detection: a systematic review highlighting age as a key confounder
Frontiers in Oncology
Post
Bayesian & maths
Positive Definite Matrix
Positive definite matrices carry the same interpretation of being ‘positive’ as a real positive number and this post will show how the idea of ‘positiveness’ extends from a…
Project
Learning analytics
Collaboration prediction in real-time
Developed an app to allow teachers to create and monitor group activities in their classroom. 🏆 Best Demo Award at LAK'23 Confer…
Paper
Bayesian & maths
Supporting Student-Centered Learning with Flexible Learning Trajectories and Open Learner Models
15th International Conference on Learning Analytics & Knowledge
Post
Bayesian & maths
Spectral Decomposition
Spectral decomsition is a method to factorize a square matrix (i.e., matrix with same nuber of rows as columns) into simpler or special matrices which facilitates efficient…
Project
Learning analytics
Impact of different temporal lenghts on collaboration quality prediction
Investigated impact of different temporal lengths on the performance of collaboration prediction model
Paper
Learning analytics
Teacher-AI complementarity: From design to implementation and reflection
15th International Conference on Learning Analytics and Knowledge
Post
Bayesian & maths
Understanding Conditional Independence
Conditional independence is a crucial building block of probabilistic graphical models (e.g., Bayesian networks). Thus, having a sound understanding of this concept is a…
Project
Data & visualization
Stakeholder-based sustainable energy scenarios
Developed an interactive map-based platform that lets users design and compare sustainable energy transition scenarios.
Paper
Learning analytics
Bringing Collaborative Analytics using Multimodal Data to the Masses: Evaluation and Design Guidelines for Developing a MMLA System for Research and Teaching Practices in CSCL
14th International Learning Analytics and Knowledge Conference
Post
Bayesian & maths
Introduction to Probabilistic Graphical Models
In this post, I will give an introduction to Probabilistic Graphical Models in simple terms. By the end of this post, you will get an understanding of what PGM is, why…
Project
Data & visualization
Group Conversation Analytics using Raspberry Pi
Developed a IoT prototype to process audio data and visualize group speaking analytics through an interactive dashboard
Paper
Learning analytics
How well do collaboration quality estimation models generalize across authentic school contexts
British Journal of Educational Technology
Post
Genomics & health
Fetching taxonomic information for any micro-bacterium using Python
In a recent project, I needed to devise a way to harmonize different gut-based bio-markers for colorectal cancer found in the literature. Since studies have reported their…
Project
Data & visualization
Literature Analytics and Visualization
Developed a library to process literature review dataset, collected at Harvard University, to visualize key insights
Paper
Learning analytics
Impact of data noise on the performance of supervised machine learning models using multimodal data to estimate collaboration quality
Journal of Learning Analytics
Post
Genomics & health
Generating and Performing Automated Test for Snakemake Workflow
Snakemake is a workflow manager which simplifies the execution of complex pipelines, but as a workflow grows, testing becomes critical. Automated tests help catch errors…
Project
Learning analytics
Moodle plugin to track video watching behavior
Developed a plugin for Moodle LMS to track users' video watching behavior using YouTube API
Paper
Learning analytics
CIMLA: A Modular and Modifiable Data Preparation, Organization, and Fusion Infrastructure to Partially Support the Development of Context-aware MMLA Solutions.
Journal of Universal Computer Science
Post
Genomics & health
Introduction to Data Anonymization
You most likely have heard the term “data anonymization” if you are involved with data directly or indirectly. In this post, we will learn about this crucial part of the…
Paper
Learning analytics
Exploring indicators for collaboration quality and its dimensions in classroom settings using multimodal learning analytics
In the 18th European Conference on Technology Enhanced Learning (EC-TEL)
Post
Genomics & health
Writing your first Snakemake rule: A quick startup guide for Snakemake
A workflow is a series of tasks or programs executed in a specific order to achieve a goal. To automate the execution of these tasks, we use a workflow manager. This post…
Paper
Learning analytics
How to build more generalizable models for collaboration quality? Lessons learned from exploring multi-contexts audio-log datasets using Multimodal Learning Analytics
13th International Learning Analytics and Knowledge Conference
Post
Bayesian & maths
A primer on discrete data distribution fundamentals
This post offers an introduction to discrete data distributions which are essential to know for every data scientist. The post starts by explaining what does it mean by data…
Paper
Learning analytics
Impact of window size on the generalizability of collaboration quality estimation models developed using multimodal learning analytics.
13th International Learning Analytics and Knowledge Conference
Post
Data & visualization
How to Successfully host your Dash application on Render
This post will provide detailed instructions on the hosting process of a dash application on Render. Render is a hosting service provider. It offers a free tier that can be…
Paper
Learning analytics
Multimodal Learning Analytics Research in the Wild: Challenges and Their Potential Solutions.
13th International Learning Analytics and Knowledge Conference
Post
NLP & LLMs
Guide to train word embeddings from scratch
This blog post explains the procedure to train word embeddings from scratch. Word embeddings are dense vector representation of text data (Check this blog pos for the basics).
Paper
Learning analytics
Studying teacher withitness in the wild: comparing a mirroring and an alerting & guiding dashboard for collaborative learning
International Journal of Computer-Supported Collaborative Learning
Post
NLP & LLMs
Introduction to word embeddings with hands-on exercises
A word embedding is a numeric representation of text in natural language processing. The representation is in the form of a vector where each number represents a dimension…
Paper
Learning analytics
Towards a partnership of teachers and intelligent learning technology: A systematic literature review of model‐based learning analytics
International Journal of Computer-Assisted Learning
Post
NLP & LLMs
Building a surname classifier using PyTorch
This post presents an NLP (Natural language processing) example based on the book NLP with PyTorch:Building intelligent language applications using deep learning. In the…
Paper
Learning analytics
Co-designing a multimodal dashboard for collaborative analytics
15th International Conference on Computer-Supported Collaborative Learning (CSCL)
Post
NLP & LLMs
Hands-on experience with `llama2` with Python: Building a simple Q&A app
This post shares the step-by-step process of running the LLAMA2 model locally on MAC. The post is a reflection of my learning of Large Language Models and their practical…
Paper
Learning analytics
Do teachers find dashboards trustworthy, actionable and useful? A vignette study using a logs and audio dashboard
Technology, Knowledge and Learning
Post
NLP & LLMs
Sentiment prediction using Pytorch
This year, I started exploring NLP (natural language processing) given its huge potential for building intelligent systems capable of understanding and generating text in…
Paper
Learning analytics
EFAR-MMLA: An evaluation framework to assess and report generalizability of machine learning models in MMLA
Sensors
Post
NLP & LLMs
Basics of Natural Language Processing in Python
Natural language processing is an interdisciplinary field combining computer science, artificial intelligence, and linguistics. Natural language processing focuses on the…
Paper
Learning analytics
Teacher AI-Supported Pedagogical Actions in Collaborative Learning Coregulation: A Wizard-of-Oz Study
Frontiers in Education
Post
NLP & LLMs
Introduction to Regular Expressions
This post will explain the basics of regular expression. Regular expression (regexp in short) is a powerful technique of text searching and text manipulation. If you are…
Paper
Learning analytics
Exploring the triangulation of dimensionality reduction when interpreting multimodal learning data from authentic settings
14th European Conference on Technology Enhanced Learning (EC-TEL)
Post
Data & visualization
Location data visualization on Map using Python
This tutorial will show you how to work with geospatial data using Python with library geoPandas. For this exercise, I have used Boston Bike Sharing dataset which is…
Paper
Learning analytics
Quantifying collaboration quality in face-to-face classroom settings using MMLA
International Conference on Collaboration Technologies and Social Computing (CollabTech)
Post
Data & visualization
Introduction to Python Dash Framework
Dashboards play a crucial role in conveying useful and actionable information from collected data regardless of context. As with economically feasible sensors, improved…
Post
Data & visualization
Computing inter-rater aggrement scores using Python
Recently, I was involved in some annotation processes involving two coders and I needed to compute inter-rater reliability scores. There are multiple measures for…
Post
Data & visualization
Understanding Gramian Angular Field
In this post, I will share my learning on this topic. I have been searching for a way to use a convolutional neural network (CNN) for my time-series data and this search led…
Post
Data & visualization
R tutorial on accessing, filtering, aggregating and plotting data
This post offers a tutorial on how to access, filter, aggregate and plot the data in R. It will mainly covers following topics
Post
Data & visualization
Setting up MQTT server and Client
This post offers an introduction to the MQTT (Message Queuing Telemetry Transport) protocol [1] and also demonstrates its usage with an example in Python (Just for info…
Post
Bayesian & maths
Statistical tests in R
This post delves into inferential statistics and explains two different statistical test. Additionally, this post also talks about the need of performing these tests along…
Post
Data & visualization
Introduction to Python's Pandas API
Pandas is a Python API for processing data in a easy and efficient way. This post offers an introduction to this amazing API , especially for beginners.
Post
Bayesian & maths
Understanding Principal Component Analysis under the hood
Let’s assume that you have a dataset with a higher number of attributes and you are thinking Is there any way to compress this information into a smaller number of attributes.
Post
Learning analytics
Facial feature extraction using OpenFace
In this post, I will discuss the work I have been doing recently. I needed to extract facial features from the recorded video and for this task, I decided to use OpenFace…
No matching items
Recognition
Best Demo Award, LAK ’23
Plus a Best Paper nomination at LAK ’23, the Estonian Youth Board Special Research Award, and EC-TEL and Dora Plus scholarships.
All awards
Path
2024 →Metrosert ASData Scientist
2018 – 2024Tallinn UniversityPhD · Developer / Researcher
2022 – 2023Harvard · Bergen · ValladolidVisiting researcher
2010 – 2012MNIT JaipurMTech