Karen is currently unavailable
Karen Danielyan
Verified Expert in Engineering
Data Scientist and Developer
Tallinn, Estonia
Toptal member since July 6, 2022
Bio
Karen is an experienced data scientist and machine learning engineer with a solid background in mathematics and statistics and a proven work track record in the fintech and ad-tech industries. Throughout his career, he has embraced different projects, including building and deploying various models for risk scoring and advertisement-exchange optimization pipelines. Karen is keen on handling big data and machine learning pipelines from data integration to final model deployment.
Portfolio
NIC MAP Vision LLC
Python, Data Science, SQL, Data Analysis, Machine Learning, NumPy, TensorFlow...
First Community Bank - Main
Machine Learning, Python, SQL, Microsoft SQL Server, Credit Risk
Tadpull, Inc.
Python, Azure, Real-time Data, Time Series Analysis
Experience
- Statistics - 8 years
- Python - 6 years
- SQL - 6 years
- Machine Learning - 5 years
- Data Science - 5 years
- Deep Learning - 4 years
- Neural Networks - 4 years
- OpenAI GPT-3 API - 1 year
Work Experience
Data Scientist | ML Engineer
2022 - 2024
NIC MAP Vision LLC
- Created a regression model to predict senior housing trends, then packaged this into an API.
- Developed a custom KNN model with tailored distance metrics, including geographic calculations. Did custom implementation on top of NumPy to handle and process large datasets for precise outcomes efficiently.
- Enhanced a TensorFlow-based NLP model and built integrations with OpenAI GPT and Embedding APIs for new use cases.
Technologies: Python, Data Science, SQL, Data Analysis, Machine Learning, NumPy, TensorFlow, PyTorch, Pandas, Jupyter, Docker, Git, OpenAI GPT-4 API, OpenAI GPT-3 API, LangChain, Embeddings from Language Models (ELMo)
Machine Learning Developer
2023 - 2023
First Community Bank - Main
- Developed a machine learning pipeline for Credit Risk.
- Performed customer segmentation based on their default risk and profitability.
- Performed ad-hoc analysis on the Credit Risk features and profitability data.
Technologies: Machine Learning, Python, SQL, Microsoft SQL Server, Credit Risk
Data Scientist
2022 - 2023
Tadpull, Inc.
- Developed and implemented anomaly detection models, accounting for historical data and seasonal traffic changes, resulting in real-time alerts for ad spend, return on ad spend, product sell-thru rate, and high-traffic pages with low transaction rates.
- Created propensity models to predict customer purchases, repurchases, and churn using individual attributes/actions (e.g., past purchases, location, website behavior), enhancing targeted marketing efforts.
- Created forecasting models to accurately predict website traffic, revenue, and projected sellout dates, considering seasonality and enabling better inventory management and marketing strategies.
Technologies: Python, Azure, Real-time Data, Time Series Analysis
Machine Learning Engineer
2022 - 2022
Verve Group
- Maintained a framework for utilizing daily data of 120 billion auctions from various partner companies and associated demand and supply partners.
- Built a framework to train traffic-shaping and bid-optimizer models and deploy them on the ad-server hourly.
- Conducted ad-hoc performance analysis, gathering insights from a vast dataset.
Technologies: Python, Spark ML, Flask, GitHub, Apache Airflow, Docker, Kubernetes, Machine Learning, Big Data Architecture, Redis, Amazon Web Services (AWS), PyCharm, Linux, Statistical Modeling, Data Science, SQL, Git, Algorithms, Jupyter Notebook, Data Analysis, Data Visualization, Machine Learning Operations (MLOps), Data Mining, Jupyter
Data Scientist
2019 - 2022
Bondora
- Built, deployed, and maintained a machine learning pipeline from data integration to a Docker image and API service in production.
- Performed ad-hoc data mining, hypothesis testing, and insight gathering for management decision-making.
- Analyzed A/B testing results using statistical methods.
Technologies: Python, SQL, Spark ML, Dask, BentoML, GitLab CI/CD, Docker, Kubernetes, Git, Machine Learning, Data Science, Statistical Modeling, Artificial Intelligence (AI), Natural Language Toolkit (NLTK), Natural Language Processing (NLP), Generative Pre-trained Transformers (GPT), Amazon Web Services (AWS), Spark, Financial Modeling, Mathematics, PyCharm, Neural Networks, Deep Learning, Statistics, Algorithms, Jupyter Notebook, Data Analysis, A/B Testing, Data Visualization, Probability Theory, Machine Learning Operations (MLOps), Time Series, Time Series Analysis, Data Mining, Data Modeling, Data Reporting, Databricks, Jupyter
Data Scientist
2020 - 2020
Deloitte Audit Analytics
- Assisted in developing ranking algorithms for companies based on random forest and clustering.
- Performed data migration from Azure to GCP, created a sustainable database structure, data processing pipelines, and ETL, and developed analytical tools.
- Gathered ad-hoc analytical insights from collected data.
Technologies: Azure, Python, Google Cloud, SQL, Machine Learning, Data Science, Data Analytics, Google BigQuery, Financial Modeling, PyCharm, Statistical Modeling, Statistics, Stochastic Modeling, Derivative Pricing, Git, GitHub, Jupyter Notebook, Data Analysis, Data Visualization, Data Mining, Data Modeling, Data Reporting, Jupyter
Data Specialist
2017 - 2019
One Market Data
- Performed tick data collection, processing, and storing for equity and futures markets and created a final sellable product with the Quant Data team.
- Covered the data processing for more than 200 exchanges.
- Developed various SQL stored procedures and functions for managing a constant data flow.
- Supported the normalization of the very granular exchange data as well as processing to fit the clients' needs like back testing of trading strategies, using the composite exchange prices for validation, order book creation, and more.
Technologies: MySQL, Linux, Shell, Python, Perl, Financial Modeling, Big Data Architecture, SQL, Git, GitHub, Algorithms, Trading, Jupyter Notebook, Data Analysis, Data Visualization, Data Mining, Jupyter
Project History
Credit Risk Modeling Pipeline
An API solution that scores loan applicants' credibility based on several data sources, including user input variables, credit bureau data, and bank statement transactions. The latter part contains an NLP model to predict expense and income categories based on the plain bank transaction descriptions. The project contains the initial data ETL, feature engineering, modeling, evaluation, and deployment.
Traffic Shaping Framework
The project is designed to shape the ad traffic on an auction level. The demand and supply sides trade ads on online auctions that happen around 120 billion times a day. The project uses machine learning algorithms and data processing tools to determine the most probable bidders for the ad supply side and not waste traffic on the cloud by sending bid requests to improbable bidders given the ad characteristics. The project is developed mainly in Python.
Sustainable Finance Project
The project aims to find a connection between global companies' carbon footprint and market risk. The Climate Finance or Sustainable Finance project includes a compound data pipeline, which uses several databases from the partner companies, extracts the relevant data, transforms, and loads in a Google BigQuery. The second part of the project provides a decision-making platform that leverages machine learning models to find a relation and causation between several ESG parameters of the companies and their balance sheets and cash flow statements. Thus, eventually calculates the market risk for those companies.
Customer Segmentation for Targeted Marketing Strategies
Developed a tree-based segmentation pipeline to optimize marketing efforts by identifying and targeting high-value customer segments. This project involved extensive feature engineering, implementing a grid search, and creating a reusable pipeline. The result was a highly effective tree traversal system that pinpointed the best and worst customer segments for retargeting, enabling the company to allocate resources more efficiently and drive higher returns on marketing investments.
Education
2018 - 2020
Master's Degree in Mathematics and Statistics
University of Tartu - Tartu, Estonia
2012 - 2018
Bachelor's Degree in Mathematics and Statistics
Yerevan State University - Yerevan, Armenia
Skills
Libraries/APIs
Scikit-learn, Pandas, NumPy, Keras, Spark ML, Dask, BentoML, Natural Language Toolkit (NLTK), TensorFlow, PyTorch
Tools
PyCharm, SQL Management Studio, Jupyter, MATLAB, GitLab CI/CD, Git, GitHub, Apache Airflow, Shell
Languages
Python, SQL, R, Perl
Paradigms
ETL
Platforms
Jupyter Notebook, Linux, Docker, Kubernetes, Azure, Windows, MacOS, Google Cloud Platform (GCP), Amazon Web Services (AWS), Databricks
Storage
MySQL, Redis, Google Cloud, Microsoft SQL Server
Frameworks
Flask, Spark
Other
Machine Learning, Statistical Modeling, Statistics, Data Science, Probability Theory, Predictive Modeling, Regression, Mathematics, Deep Learning, Stochastic Modeling, Derivative Pricing, Calculus, Linear Algebra, Analytical Geometry, Differential Equations, Time Series Analysis, Stochastic Differential Equations, Data Analytics, Artificial Intelligence (AI), Financial Modeling, Data Analysis, A/B Testing, Machine Learning Operations (MLOps), Time Series, Data Mining, Data Modeling, Data Reporting, OpenAI GPT-3 API, Neural Networks, Big Data Architecture, Cloud Computing, Option Pricing, Natural Language Processing (NLP), Optimization, Google BigQuery, Algorithms, Trading, Data Visualization, Generative Pre-trained Transformers (GPT), Credit Risk, Real-time Data, OpenAI GPT-4 API, LangChain, Embeddings from Language Models (ELMo)
Collaboration That Works
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
1
Share your needs
Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2
Choose your talent
Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3
Start your risk-free talent trial
Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.
Top talent is in high demand.
Start hiring