Experience

Huma
• Primary and Secondary Education
• Senior Back End Engineer

  • Jun 2021 - Present


Communere Ltd.

• United Kingdom
• Software Development
• 1 - 100 Employee

• Back-end Developer Python

  • Jan 2019 - May 2021

  Design, implement and maintain core algorithms for Agxio. My first duty was designing and implementing an image stitching algorithm which creates a mosaic image from sets of thousands of aerial imagery. It is done by designing a cutting edge image processing algorithm which is built on OpenCV functions. The image processing flow is designed to run concurrently using full multicore capacity of the processor. Implemented several web APIs using flask and connexion with OpenAPI (Swagger) specification to facilitate access to the server. Developed a data science dashboard which provides exploratory data analysis (EDA), dataset manipulation and fitting classification, regression, and image classification models. Dataset manipulation includes applying standard-scaler, handling categorical values, etc. The classification models include MLP, decision tree, Adaboost, Gaussian process, Bayesian network, SVM (different kernels), random forest, etc. Developing a flexible model creation API, I enabled the users to create arbitrary Deep Learning models using Keras and TensorFlow and use those for image classification. Designed and implemented a scalable multi-worker architecture for long time processes. This architecture helped users to use their specific resources and also gave them online resource reports using SocketIO. My tech. stack is Python (Pandas, seaborn, matplotlib, scikitlearn, tensorflow, SocketIO, Flask, OpenCV,...), MySQL, SQLite, and C++. Achievements * Designed and implemented a concurrent image stitching engine using OpenCV. The algorithm is designed based on homography calculation, since open-source stitching algorithms are not able to deal with large number of images * Implemented a cross-platform image uploader using Qt (works on Linux, Windows and Mac) * Developed a data science dashboard which provides exploratory data analysis, dataset manipulation and fitting different model types including image classification. Show less



White Fox Cloud

• Australia
• Software Development
• 1 - 100 Employee

• Back-end Developer Python

  • Mar 2020 - Feb 2021

  Designed and Implemented a PDF rasterization/tiling tool using Amazon AWS Lambda Functions. The final solution reads and writes the PDF file and the results from/to an S3 bucket. In order to process the PDF file fast, I used PyMuPDF package. First created a native lambda function (required packaging), then did the project again using serverless framework. Achievements • Decreased the tile creation process from 1500 seconds to less than 20 seconds Designed and Implemented a PDF rasterization/tiling tool using Amazon AWS Lambda Functions. The final solution reads and writes the PDF file and the results from/to an S3 bucket. In order to process the PDF file fast, I used PyMuPDF package. First created a native lambda function (required packaging), then did the project again using serverless framework. Achievements • Decreased the tile creation process from 1500 seconds to less than 20 seconds



Zharfpouyan R&D Co.

• Developer

  • Mar 2016 - Jan 2019

  working on a BSD based embedded network device. In addition to work with a creative team of skilled C/C++/Python programmers, I was responsible for maintaining the git repository. The task of our team was to maintain a huge codebase, consists of several BSD and Linux based projects. These projects include routing daemons, our changes in the operating system kernel and our testing tools. Clustering the hourly web traffic based on the web preferences. The aim of the project was to enable the web caches to keep only the necessary web pages. I implemented both k-means and decision tree. It was implemented using Python, C++ and mysql. We moved from SVN, taskfreak and mantis to git and gitlab. I was developing with C and C++. Achievements * Improved a previously designed web cache using machine learning algorithms * Improved the image system, which was an information storage and retrieval responsible for generating new systems * Designed and developed a new protocol to keep network appliances’ software up to date. * Gained good experience in Linux, Unix based systems Show less

• Programmer and System Designer (remote)

  • Jan 2013 - Mar 2016

  Designed, implemented and maintained the core algorithm for the Visography TRACC. TRACC is a software package to address all aspects of travel time fully and cost mapping using digital road networks, public transport timetable data and flexible, on-demand transport. My first duty was increasing the speed of a previously developed algorithm in C#.NET. It was done by proposing and implementing a new method for importing road-networks into a graph, redesigning and producing a fast algorithm for route finding within that graph. It also included a fast and accurate solution for public transport route finding. I implemented numerous unit tests to check the work too. I also worked on a multiple traveling salesman problem with load and distance constraints: this project was a package distribution planning core. It helps to transfer more than 500 packages using more than ten electric cars. The problem was solved considering the electric cars' constraints, i.e. road slop, speed and the range. Besides, I have implemented an extra package (beyond the constraints) removal algorithm. To solve the problem, I implemented a multi-threaded genetic algorithm. Most challenging part of the design was crossover and fitness functions because the response time of the algorithm was critical. The final product produced robust results and converges fast. We used Scrum Agile Methodology, and Jira Software as the issue tracker, previously SVN, and then git for versioning. The development was done using C#.NET. Achievements * Improved the performance of a previously designed routing algorithm * Designed and implemented a multi-destination routing algorithm * Designed and developed a public transportation routing system * Designed and developed a package distribution planning solution for electric cars using genetic algorithm * These routing engines are still in use Show less