Mert Bakır

Senior Data Scientist

Experience

Evreka

Principal Data Scientist 08.2022 - Present
  • Developed a machine learning model and deployed into production, to classify time series data with 87% F1-score significantly aiding the company in identifying faulty sensors.
  • Developed a machine learning model and deployed into production, to forecast the fill level of waste containers, enabling the company to optimize waste collection routes.
  • Developed algorithms to mitigate noise in sensor data and improved the event detection, achieving a reduction of false positives by 20% and improving quality of downstream analytics.
  • Mentored and trained junior data scientists and engineers, providing guidance and support in their professional development.
Data Scientist/Engineer 10.2020 - 08.2022
  • Developed Route Optimization service from the ground up including heuristic algorithms, backend endpoints, REST API integrations with third-party map services. The service is designed to address various scenarios, including solving Capacitated Vehicle Routing problems with multiple refills, or any other subset of the problem.
  • Developed and implemented a tailored Capacitated Clustering Algorithm to generate solutions for high-scale Vehicle Routing Problems such as 10,000 nodes.
  • Formulated and implemented a Mixed Integer Programming model for a Shift Scheduling Problem using Python and the CPLEX API.
  • Conducted analyses for waste collection companies, focusing on capacity planning. Successfully reduced fleet sizes by up to 17% by providing optimized work plans.
  • Set up and maintained open-source map engines, eliminating costs associated with proprietary map services.
  • Built the data infrastructure, ELT/ETL pipelines using mainly open-source technologies like Dbt, Airflow, Cosmos, Mongodb, Postgresql, AWS, in the absence of pre-existing infrastructure.
  • Designed and built a data warehouse, incorporating best practices in dimensional modelling. The warehouse serves both internal reports and customers seeking access to their data directly.
  • Delivered impactful insights to customers through the implementation of PowerBI Embedded, solidifying it as a core feature of the product.
  • Empowered the hardware team by crafting dashboards for monitoring the health of thousands of IoT devices.

Education

Yıldız Technical University

Msc. Industrial Engineering
2019 - 2023 | GPA: 3.71
Master's Thesis

Gazi University

Bsc. Industrial Engineering
2013 - 2017 | GPA: 3.35

Skills

  • Python - FastAPI - Flask
  • R - Bash - Powershell
  • numpy - sklearn - sktime
  • catboost - lightgbm - mlforecast
  • pandas - geopandas
  • SQL - Postgis - Postgresql
  • Mongodb - Redis
  • Influxdb - Kapacitor
  • Airflow - Dbt - Cosmos
  • PowerBI - Grafana - Superset
  • CPLEX - OR-Tools
  • OSRM - Valhalla
  • AWS - Linux - Docker
  • Graylog - Loki - NGINX
  • Awards

    Tübitak - Graduation Project

    A Decision Support System for 3D Cutting Problem - 2017

    Languages

    Turkish - Native

    English - Professional

    Mert Bakır

    Senior Data Scientist

    Experience

    Turkish Technic

    Observability Team Lead 06.2019 - 10.2020
    • Pioneered a monitoring project that introduced comprehensive observability, significantly enhancing insight into IT operations where none existed previously.
    • Built streaming data pipelines to capture metrics and logs from IT services, infrastructure, and diverse IoT devices within the Data Center.
    • Developed Python and Kapacitor scripts for processing time series data and generate alerts in case of anomalies.
    • Crafted versatile and customizable real-time dashboards using Grafana to enhance operational efficiency for IT engineering teams.
    System Engineer 10.2018 - 06.2019
    • Managed server infrastructure, ensuring operational systems for other IT departments.
    • Automated various administration tasks on Linux and Windows systems using Bash and Powershell, contributing in team efficiency.

    Turkish Aerospace Industries

    Operations Research Intern 11.2016 - 07.2017

    Created a software solution with the objective of minimizing raw material waste in a specific cutting process of metal blocks. Formulated and implemented a constructive heuristic algorithm that generates optimal cutting patterns, complemented by a meta-heuristic search to improve the initial solution. Achieved a notable 35-48% reduction in material waste. Developed an additional algorithm that backtracks the generated cutting patterns and translates the results into precise instructions for the human operator. These instructions are supported with 3D visualizations representing each cutting step.