Edward is a Senior Python Developer with 6 years of experience creating back-end services using Flask, Django, SQL, MongoDB, Redis, AWS, and Circle CI. He is an AWS Certified Developer with several prestigious technology awards and additional skills in C++, JavaScript, Docker, Kubernetes, Git, GitHub, OOP, and Design Patterns. Edward is also keenly interested in AI, space, and gene editing and works with clients and distributed teams in a fully remote capacity.
Developed and maintained the app's back end, allowing users to watch and manage retail inventories.
Created three integrations with third-party back-end services to ingest automatically, process automatically, and store necessary data daily and implemented redundancy to cope with regularly failing third-party services.
Implemented performant SQL queries to analyze and process 5 million rows with complex joins, returning summarized results to user queries in real time.
Migrated a large portion of the project from Python 2 to Python 3.
Fixed unit and functional tests caused by Python 2 to Python 3 migration.
Handled most of the CI/CD (Circle CI) changes required when transitioning from Python 2 to Python 3, ensuring it worked well with the existing CI/CD systems.
Developed and maintained the app's back end that allowed users to search, track, and log hundreds of multiphase experiments with dynamically changing specifications.
Connected the app to Slack and Asana by using their APIs, created bots to notify app users about their tasks and overall progress, and synchronized the data between the database, Slack, and Asana.
Developed the CI/CD that ran tests and database migrations and deployed the back end to AWS Lambda.
Designed and developed the "dataset format for AI" and played a key role in getting the project from 0 to 2,400 stars (github.com/activeloopai/hub).
Co-developed the app back end that visualizes AI datasets and allows users to zoom in/out, batch, and more.
Developed a pipeline for ingesting datasets, training models, and running inference in a multimachine, parallelized environment using Kubernetes-like technology.
Developed a C++ back end, co-developed a Python back end, and took part in all stages of project development.
Designed and developed a custom, real-time database for aerial heatmaps with zoom-in-and-out support (the database and its Python back end allowed users to view 80GB of data in real time despite strict I/O limitations).
Modified an open-source DSD project, created a cross-language interoperability layer between LabVIEW and C++, allowing LabVIEW developers to access the DSD functionality (the original project is available on github.com/szechyjs/dsd).
Sudoku Image to Text Parser was a university project aimed at converting Sudoku photos into digital form, using OpenCV and open-source algorithms to extract the Sudoku grid.
The project employed a breadth-first search (flood fill) technique to separate each digit from the grid and an optical character reader labeled each digit image.
The optical character reader was trained using the MNIST dataset and a known convolution network which was further fine-tuned using programmatically generated (Pillow) printed digits. The OCR achieved 99% accuracy, translating to 90% accuracy for the entire Sudoku grid.
Research project focuses on detecting drone or plane's location by tracking changes in surface images beneath the vehicle, acting as a backup if GPS navigation fails.
Utilizes linear algebra and OpenCV algorithms for position detection which, although not precise, is enough to guide the drone to the desired place.
The performance of the analysis is efficient enough to run on an onboard Raspberry Pi.
