Merve Acar
Verified Expert in Engineering
Data Scientist and Software Developer
San Francisco, CA, United States
Toptal member since March 11, 2019
Merve is an experienced machine learning engineer who takes pleasure in revealing the story of data and building predictive models. She has a proven track record of designing and implementing pipelines for extracting, validating, cleaning, transforming, and modeling data. Merve is passionate about solving real-world industry problems and is eager to take on new challenges and opportunities.
Portfolio
Experience
- MySQL - 7 years
- Python - 6 years
- Machine Learning - 5 years
- Data Mining - 5 years
- Data Science - 5 years
- PyTorch - 4 years
- Deep Learning - 4 years
- Tableau - 2 years
Availability
Preferred Environment
Git, PyCharm, Jupyter Notebook, Linux, Windows, Amazon EC2, Jira, Slack
The most amazing...
...automated machine learning tool I've developed leverages the meta-learning power to select the most optimal algorithm and its parameters, adapting to any task.
Work Experience
Senior Data Scientist
XyloRisk Inc.
- Developed and validated risk scoring models by leveraging machine learning techniques, historical damage data, and external dataset comparisons to improve risk assessment accuracy.
- Implemented feature impact analysis to evaluate individual vs. interaction effects on risk scores, refining predictive modeling approaches beyond manual thresholding.
- Conducted fire perimeter and WRC dataset analysis to assess how proximity to past wildfires influences risk scoring and identify discrepancies in external risk classifications.
Senior Data Scientist | Data Engineer
Clara Foods Co DBA The EVERY Company
- Conducted statistical analysis and predictive modeling on fermentation data, contributing to the successful forecasting of titer values for product development.
- Optimized experimental protocols by analyzing protein content and texture scores, contributing to improved product formulation.
- Led the migration of Excel calculations to a Python-based back end, boosting efficiency by 30%, and optimized API endpoints for user input and file uploads, enhancing data processing speed and user experience.
Data Analyst
Trust & Safety Laboratory
- Automated the collection of social media accounts spreading misinformation from multiple fact-check sites.
- Utilized Hugging Face's CLIP model for zero-shot learning to detect harmful content in images.
- Performed analyses to visualize bad actors and their friends of friends network graphs.
- Translated client requirements into interactive dashboards in Tableau.
Data Scientist
Turkish Aerospace Industries
- Developed a dashboard-based surveillance system to improve a factory's work processes using IP camera recordings. Applied video and image processing algorithms using the OpenCV library together with object detection and object tracking algorithms.
- Built an LSTM-based model to identify people's actions and improve work processes in a factory.
- Developed a predictive maintenance model using ARIMA and LSTM algorithms which provides insight into a plane part's breakdown using the time-series data of a plane component. Applied data manipulation, analysis, and visualization.
Machine Learning Engineer
Vitus Commodities
- Took part in several data scraping projects using Selenium, API calls, the requests library, and more.
- Created reports on Microsoft Power BI for data visualization.
- Implemented a multilayer perceptron model using Python and Keras to forecast the natural gas demand in the UK for the coming days.
- Deployed an LSTM model that predicts Turkey's electricity price for the next few days.
- Implemented a scraper to obtain and manipulate GFS weather data to use as a source for model training.
- Investigated deep learning methods to enhance the performances of the current working models for time-series data.
- Implemented an outlier detection project consisting of probabilistic and clustering-based algorithms and an autoencoder method to detect extreme days concerning the UK's natural gas demand.
Machine Learning Engineer
Independent Work
- Implemented data preprocessing, data imputation, feature extraction, and model creation modules for the Vitriol project using Scala.
- Researched and tested a meta-learning strategy to predict the best model with the best parameters for a given problem using Scala and Spark.
- Implemented a parser to handle unstructured data that comes from different sources using Python.
- Worked with big data using Apache Spark framework and Scala.
Software Developer
C3S Command Control & Cybernetic Systems
- Developed connector reliability testing software that controls the connection between PCI cards and connectors on the Linux platform.
- Built software that calculates how much time an employee spends at the office.
- Wrote SQL database queries to analyze an employee's working schedule.
Software Test Developer
Taleworlds Entertainment
- Developed automated tests for Mount&Blade: Bannerlord II project.
- Monitored test results and reported bugs found in prerelease software on a daily basis.
- Performed unit tests and integration tests to determine if the game scenes were working correctly.
- Worked within an Agile environment with multiple teams.
Software Developer
TUBITAK | The Scientific and Technological Research Council of Turkey
- Developed a parental control tool for Pardus, a Linux distribution supported by the Turkish government.
- Implemented content filter, usage control, and monitoring modules.
- Gained experience in open-source development and the security field.
Experience
Vitriol
http://senior.ceng.metu.edu.tr/2016/mallorn/I used a meta-learning strategy to select the most appropriate algorithm and its parameters. This project is implemented in Spark and the Scala programming language to handle big data.
Natural Gas Demand Forecasting
First, I implemented an extreme day detection module to label the data as extreme or not extreme. An oversampling method helped enhance extreme days because they were a small portion of the data. I also implemented a dynamic weighted ensemble model using a multilayer perceptron (MLP) and a linear regression model to consider both linear and non-linear trends.
Stock Price Prediction
Denoising of Images
PriceTag
Pardus Gozcu
Education
Master's Degree in Computer Engineering
Istanbul Technical University - Istanbul, Turkey
Bachelor's Degree in Computer Engineering
Middle East Technical University - Ankara, Turkey
Certifications
Using Python to Access Web Data
Coursera
Structuring Machine Learning Projects
Coursera
Convolutional Neural Networks
Coursera
Practical Time Series Analysis
Coursera
Fundamentals of Visualization with Tableau
Coursera
Google Cloud Platform Big Data and Machine Learning Fundamentals
Coursera
Neural Networks and Deep Learning
Coursera
Machine Learning Foundations: A Case Study Approach
Coursera
Skills
Libraries/APIs
Matplotlib, Scikit-learn, Pandas, PyTorch, Keras, Slack API, XGBoost, OpenCV, Natural Language Toolkit (NLTK), Spark ML, PyQt, Beautiful Soup, TensorFlow, Protobuf, X (formerly Twitter) API, NetworkX
Tools
Microsoft Power BI, PyCharm, Slack, Git, Seaborn, Plotly, Tableau, Jira, Bazel, You Only Look Once (YOLO)
Languages
Python, SQL, Python 3, Bash, Scala, C++, Haskell, Bash Script, R, XML, Snowflake, JavaScript, HTML, CSS
Frameworks
Selenium, Spark
Platforms
Windows, Linux, Jupyter Notebook, RapidMiner, Amazon EC2, Amazon Web Services (AWS), Gephi, AWS Lambda, Google Cloud Platform (GCP)
Storage
PostgreSQL, MySQL, Data Pipelines, Databases, Amazon S3 (AWS S3), JSON
Industry Expertise
Social Media
Other
Machine Learning, Data Science, Predictive Modeling, Data Processing, Web Scraping, Google Colaboratory (Colab), Regression, Classification, Decision Trees, Artificial Intelligence (AI), CSV, Exploratory Data Analysis, Data Cleaning, Computer Vision, Metric Learning, Time Series, Data Mining, Visualization, Deep Learning, Statistics, Object Detection, Object Tracking, Time Series Analysis, Data Visualization, Software Engineering, Convolutional Neural Networks (CNNs), Image Analysis, Neural Networks, Data Structures, Data Analysis, Data Modeling, Version Control Systems, Models, Modeling, Communication, Data Analytics, APIs, Data, Unsupervised Learning, Supervised Machine Learning, Decision Modeling, Data-driven Decision-making, Data Engineering, Dashboards, Reports, Data Scientist, Statistical Analysis, Remote Sensing, Natural Language Processing (NLP), Cloud Services, Design, Machine Learning Automation, Gated Recurrent Unit (GRU), Generative Adversarial Networks (GANs), Long Short-term Memory (LSTM), Data Scraping, Feature Analysis, Amazon RDS, Sentiment Analysis, Generative Pre-trained Transformers (GPT), Okta, OpenAI GPT-3 API, Full-stack Development, Spatial Statistics, GeoPandas
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring