Close

Fatih Kubilay Yavuz

Data Scientist at McKinsey & Company

Download Resume

About Me

My aim is to combine the data science and machine learning knowledge I learned by myself and the theoritical knowledge of statistics I learned from my university, and deliver efficient and fast outputs.

Skills

Experience

McKinsey & Company

Data Scientist


Geospatial Analysis and machine learning modelling,
Mathematical modelling

INFARM

Data Analyst


Looker dashboard deployments and creating data models,
B2C data centralization using Python and Airflow

Getir

Data Scientist


Pricing project for optimization of the products’ prices using ML,
Implemented generic heuristics and CVRP for courier routing,
Analysis for other departments,
Heatmap visualization of the warehouses
Campaign optimization using ML and Tableau dashboards

LC Waikiki

Research Intern

A machine learning project based on building a fashion recommendation engine from scratch was the objective of the internship. Several deep learning models were used in the course of the project which classify users' images and return most similar items through hundreds of fashion images and labels from LC Waikiki's e-commerce site. A successful working prototype was built in one month.

Responsibilities:

Data cleaning and manipulation of hundreds of thousands images and labels.
Building a custom batch processing pipeline for scaling deep learning models.
Deployment of deep learning models to a Flask application as a minimum valuable product.
Preparation of technical reports and presentations every week and final technical research paper.

Anadolu Efes Beverage Group

Business Solutions Intern

Worked on a new CRM solution’s integration to the company.

Enqura IT R&D

Business Intelligence Intern

Worked on a large-scale company’s data, conducted feature engineering, data visualization using Power BI, and SQL

Education

Bogazici University

September 2015 - June 2020

Bachelor of Science in Industrial Engineering

Ranked 553th among 2 million students in Turkish University Admission Exam 2015.

Projects and Other related activities

Teknofest Artifical Intelligence Competition

Participated with my teammate Burak with the name TigersAI. We have competed with no advisor and sponsors, and used only Google Colab and Drive. In the semi final we came the third. In the final stage, we came the fifth. The competition was about 6 months, and concluded with 5 days of final stage. The objective in the competition was to detect vehicles and pedestrians from a drone footage. Our model comprised of 3 RetinaNet models, in the final stage, we have ensembled them.

If you want to find out more about the project, feel free to contact me via LinkedIn or contact box at the bottom.

Vestel V-Intelligence Computer Vision Competition

The objective was image classification. Our objective was to determine the care symbols attached to clothes. Some image preprocessing experimented. Bunch of state of the art image classification models are used and DenseNet121 and DenseNet201 turned out to be the feasible options in the project. The model code is available on my github.

View Project

Turkish Airlines Travel Datathon

The objective was predicting the passanger count of a two legged flight. The data contained 93 million rows, so it was kind of a big data objective. We worked on a sample of the data, conducted feature engineering, used Catboost, Keras, and Linear Regression. Our point on using the LR was to create a baseline for the evaluation score.

SAP SACkathon

Objective was to create a dashboard for basketball coaches, players. The competition held online. The final stage was held in Barcelona, we were one the teams invited to Barcelona. The event was organized by SAP and Euroleague. During the final stage, we presented our projects to C-level managers of Barcelona, Sap, and Euroleague.

Yes, at the time my wrist was broken. :)

Serathonin - Capital Market Hackathon

The project was creating a machine learning backed personal dashboard for capital market investors. We used tweets and KAP news and made sentiment analysis on the highlighted sentence whether it is possitive for the selected stock or not. Then, a Reinforcement model was deployed, but decision tree model (strangely) proved to be better, so we went for that one. Then, we created a basic web app,a dashboard,using flask.

LC Waikiki Datathon

The objective was to predict whether a customer will churn in the given period. We conducted feature engineering, used Lightgbm,and Keras. Lightgbm gave a better performance.

Get in Touch