Ahmet Erdem, Developer in İstanbul, Turkey
Ahmet is available for hire
Hire Ahmet

Ahmet Erdem

Verified Expert  in Engineering

Data Scientist and Software Developer

İstanbul, Turkey

Toptal member since May 4, 2021

Expertise
Artificial IntelligenceNLPMachine LearningDeep LearningChatbot DevelopmentPythonPyTorchSparkApache AirflowGitLinuxDockerJavaScriptRoboticsC++
Bio

Ahmet is a Kaggle grandmaster with 23 gold medals in a variety of problems, such as forecasting, science, NLP, and computer vision. He made it to the top 10 in Kaggle's global ranking. Ahmet also maintains several open-source projects and enjoys working on end-to-end machine learning solutions.

Portfolio

NVIDIA
Python, PyTorch, TensorFlow, Docker, RAPIDS, Machine Learning, Deep Learning...
ING Group
Python, Spark, Docker, Apache Airflow, Model Validation, Machine Learning...
SAP
JavaScript, SQL, SAP HANA

Experience

  • Machine Learning - 5 years
  • Model Validation - 5 years
  • Git - 5 years
  • Python - 5 years
  • PyTorch - 4 years
  • Docker - 4 years
  • Deep Learning - 3 years
  • Apache Airflow - 3 years

Availability

Part-time

Preferred Environment

Jupyter, PyCharm, Git, Docker, Linux

The most amazing...

...projects I've co-developed are two open-source packages used by many data scientists for feature importance and string matching.

Work Experience

Senior Data Scientist

2020 - PRESENT
NVIDIA
  • Competed on Kaggle for testing NVIDIA products. Provided feedback and guidance on RAPIDS (open-source package for GPU-accelerated data science) and other NVIDIA solutions.
  • Shared GPU-accelerated machine learning solutions with the ML community on YouTube and Kaggle.
  • Improved GPU sales forecasting using convolutional neural networks and created an automated early warning system for data anomalies like the COVID-19 period. Set the CI/CD pipeline for the project.
Technologies: Python, PyTorch, TensorFlow, Docker, RAPIDS, Machine Learning, Deep Learning, Model Validation, Chatbots

Data Scientist

2016 - 2020
ING Group
  • Implemented customer account matching module for an internal application that can cross-match millions of accounts within an hour.
  • Implemented a private individual classification module for an internal application that could protect private individual data with 99% precision and allowed us to use 60% of the non-customer data we had.
  • Implemented a peer detection module for preventing bankruptcy within customers who are crucial to each other.
  • Developed an application with a smart and fast search for legal documents, which reduced hours of manual work to seconds.
  • Developed a secure and accurate in-house translation application for legal documents written in foreign languages.
  • Improved the search module with a question answering model, which could extract information from documents using natural language.
  • Supervised a master thesis student on genetic algorithms for regex learning.
  • Prepared technical assignments and interviewed the candidates for the fast-growing team.
Technologies: Python, Spark, Docker, Apache Airflow, Model Validation, Machine Learning, Cython, Elasticsearch, Spark ML, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP)

Software Engineer

2014 - 2015
SAP
  • Implemented a code versioning tool for SAP HANA applications.
  • Optimized aircraft maintenance schedules using predictive analytics for a customer.
  • Created a dashboard for the customer to easily analyze their own flight data.
Technologies: JavaScript, SQL, SAP HANA

Experience

LOFO Importance

https://github.com/aerdem4/lofo-importance
Leave One Feature Out (LOFO) Importance calculates the importance of a set of features based on a metric of choice, for a model of choice, by iteratively removing each feature from the set and evaluating the performance of the model, with a validation scheme of choice based on the chosen metric.

Memory-efficient Sparse Vector Matching

https://github.com/ing-bank/sparse_dot_topn
Comparing very large feature vectors and picking the best matches, in practice, often results in performing a sparse matrix multiplication followed by selecting the top-n multiplication results. In this package, we implemented a customized Cython function for this purpose. When comparing our Cythonic approach to using SciPy and NumPy functions, our approach improves the speed by about 40% and reduces memory consumption.

Education

2015 - 2016

Master's Degree in Artificial Intelligence

KU Leuven - Leuven, Belgium

2009 - 2014

Bachelor's Degree in Computer Engineering

Bogazici University - Istanbul, Turkey

Skills

Libraries/APIs

PyTorch, Spark ML, TensorFlow, RAPIDS

Tools

Apache Airflow, Jupyter, PyCharm, Git

Languages

Python, C++, JavaScript, SQL

Frameworks

Spark

Platforms

Docker, Linux, SAP HANA

Storage

Elasticsearch

Other

Machine Learning, Model Validation, Natural Language Processing (NLP), Deep Learning, Data Science, Artificial Intelligence (AI), Chatbots, Robotics, Cython, Generative Pre-trained Transformers (GPT)

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