Dan Andrei Ganea Email & Phone Number

Utrecht, Ut, Nl

Oakland, California, Us

Part or the SQL Server team. Working on high volume low latency data replication/ change data capture. We help customers build robust ETL/ELT pipelines.

Amsterdam, North Holland, Nl

Currently part of the Payments and Invoicing team, with a background in coordinating company-wide projects. Previously part of the Product Growth and Inventory management teams.- Implemented RESTful and GraphQL APIs using Kotlin, deployed using Kubernetes. Worked with an event-driven architecture where microservices communicated via AMQP.- Contributor to platform libraries E.g.: transaction propagation, retrying mechanisms, integration of our services with Redis and CosmosDB (via the MongoDB interface), using the outbox pattern to send messages.- Responsible for a service serving per-item recommendations meant to handle a large number of requests, using Redis.- Led the creation of a third-party marketing service, aggregating unsold item data and enabling diverse export formats. Achieved a 24X improvement in export speed and ensured high throughput scalability, utilizing MongoDB.- Introduced SLOs for some services and implemented load-tests using K6 and monitoring via Grafana.- Optimised latency and throughput of core functionality in our platform by profiling and optimising SQL queries Technology stack: Kotlin, KTor, Java, Docker, Kubernetes, Redis, RediSearch, CosmosDB, MongoDB, GraphQL, SQL, Grafana, Gradle, GitWe use event-driven architecture in a distributed microservice-based architecture.

Zaltbommel (Nl), Middleton, Wi (Usa), Wetzlar (Ger), Nl

- Lead the design and implementation of a data-pipeline used for detecting changes in city advertisements regions. Data was stored in Azure CosmosDB. Used: Python, FastAPI, CosmosDB, Docker.- Implemented novel machine learning techniques for detecting changes between city advertisements. Developed an evaluation framework for this task and improved previous performance by 50%. Used: Python, PyTorch.- Worked on improving face and license plate blurring, resulting in a 30% accuracy increase.

Zaltbommel (Nl), Middleton, Wi (Usa), Wetzlar (Ger), Nl

Working on my master’s thesis in Computer Vision, which relates to instance segmentation for RGBD images using machine learning and deep metric learning techniques. Have achieved state of the art results on public datasets.Used technologies: Python, PyTorch, Tensorflow.

Utrecht, Utrecht, Nl

Artificial Intelligence Teaching AssistantResponsible for teaching seminars and assisting students with projects in the Artificial Intelligence course for second year bachelor's students. Coursework includes: Search and Path Planning algorithms, Machine Learning with Neural Networks, Reinforcement Learning (Value Iteration, Q-Learning), Fuzzy Logic. Homework and seminars were based on UC Berkeley CS188 course.Technologies used: Python

Utrecht, Utrecht, Nl

Image Processing Teaching Assistant. Responsible for grading student's assignments as well as providing aid for students during live walk-in sessions and e-mail or Slack. The course covers basic techniques in image processing like histogram equalization and edge detection as well as more advanced subjects such as computing feature descriptors( SIFT , HOG ) or the Hough transform.Technologies used: C#

Stamford, Connecticut, Us

Worked in the automotive division of Harman for the BMW System Functions team. Developed and debugged new features in regards to the component monitoring and diagnostic system of the car. - Improved the speed of updating the car software (flashing) by 30% by enabling data compression.- Was a part of implementing a synchronized communication queue for multiple threads.- Was the main programmer responsible for implementing an anti-fraud module for our system.Used C++ , C, Git . Worked in an Embedded Linux environment where multithreaded programming was key.

Master'S Degree, Computer Science

Bachelor'S Degree, Computer Science