Mustafa Abdel-Hussein
I have studied Civilingenjör Teknisk Fysik—MSc Machine Learning. Currently, I am working as a Machine Learning Engineer at Synective Labs. I have a proven record of multiple projects besides school. They include, among other things, dimensionality reduction—I regained the original categories after I reduced the dimension from 784 to 2.
within NLP—I developed a heuristic method to calculate attributes to extract features from music clips to classify emotion in the song. Reinforcement Learning—I created with Python and TensorFlow an agent that plays 5 in a row. I designed every aspect, such; as the gym and environment using OpenAi, reward policy, and experimented with different Artificial Neural Networks.
I like to tinker with Jetbot, Raspberry pi, and Arduino in my spare time. I have collected data, trained, and deployed Machine Learning models on Jetbot and Arduino.
Meanwhile achieved the highest possible score in all performance surveys on service and sales in my part-time work Am skilled in balancing family time, work, and school.
Currently, I'm eager to expand my horizons and make a positive impact by taking on a hands-on position in Machine Learning.
Work Experience
Past Experience
September 2022 - Present
Machine Learning Researcher | C++ | Python
Project: PU Sensor - Embedded Software (Role: Algorithm Developer)
Objective: Transitioning from end-of-life (EOL) microcontroller units (MCUs) while integrating new sensors and functionalities.
Engineered a checksum algorithm to ensure message integrity.
Developed an I2C communication protocol facilitating interaction between the IO device and MCU.
Collaborated in a multidisciplinary team to transition from EOL MCUs to updated versions.
Actively liaised with stakeholders to refine software requirements, honing my expertise in bridging business and technical perspectives.
Technology Stack: C/C++17, Github, STM32, I2C
Project: RISE Machine Learning Research (Role: Reporting & Analytics)
Objective: Investigating the remaining useful life of TO2XX components using sparse and noisy data.
Conducted intricate data analyses and produced comprehensive reports.
Executed data scraping, labeling, and cleaning to prepare datasets for analysis.
Created and tuned custom machine learning models, assessing their performance against multiple evaluation metrics.
Integrated disparate information sources to enrich analytical insights.
Deployed inference models on both microcontrollers and cloud-based services.
Authored a Python script to automate data generation, model training, and performance evaluation through YAML configuration.
Utilized Azure Cloud services for remote sensor data logging and analysis.
Technology Stack: Python, C/C++17, TensorFlow, PyTorch, Scikit-Learn, Pandas, Github, Bitbucket
Nov 2021 - Jun 2022
Junior Software Engineer
I worked with a large Swedish telecom company with their flagship project Cloud RAN, as a developer.
In a short timespan, I managed to;
Script that lookup acronym and returned the result.
One drawback of the script was that one acronym could return multiple results. Therefore a proposal for implementing an NLP model to order the outcome based on the context given by the user, which I am developing with two other talented ML engineers from the office.
Improve team dynamic and cooperation.
Achieve a fundamental understanding of Cloud Ran and telecom, e.g,
Central Control Unit Control Plane CU-CP
Central Control Unit User Plane CU-UP
Propose multiple ideas; the company implemented numerous of them; e.g.,
Hackaton.
A machine learning moment in Office Olympic highlights the importance of quality in data collected
Skills
Python
Excellent
I have a solid knowledge of Python, also, its limitations in terms of speed or when it is preferable to rewrite your code using the Numba library.
TensorFlow
Very Good
My go-to deep learning library when I need more control than Keras can offer and when I want to deploy a Machine Learning model on embedded devices or microcontrollers such as Arduino and Jetson.
C++
Very Good
Knowing C++ made it easier to work with Arduino and to learn new languages. Understanding the pitfalls such as memory leaks, pointers, and references have developed my skills in other coding languages such as Python.
PyTorch
Good
PyTorch is the most common framework used in research articles, therefore having a solid understanding facilitates the reimplementation of research articles when going through its code.
NLP
Very Good
Have a solid understanding of how attributes affect features extracted. Has developed a heuristic method for extracting features from music clips. You can read about it in this post.
SQL
Good
I have enrolled in an online course. Currently are familiar with the manipulation, queries syntaxes, aggregate functions, and multiple tables, but not for so long!
Keras
Very Good
Perfect for prototyping and experimenting with minimum code during a project or idea. I usually start here and transit towards PyTorch or TensorFlow when I need more control.
Academic Experience
2017- 2021
Master of Science: Machine Learning, KTH Royal Institute of Technology
Courses: Speech and Speaker Recognition, Visualization, Artificial
Neural Networks and Deep Architectures, Ethical Hacking, Scalable
Machine Learning and Deep Learning, Introduction to Robotics,
Deep Learning Advanced Course, Artificial Intelligence (Master's thesis left)
2021
Course: Artificial Intelligence For Game Programming 2 C++, Uppsala Universitet
Received highest grade.
Build an Ai gym environment with OpenAi and trained an agent to play 5 in a row, using TensorFlow.
Everything was build from scratch.
2012-2015
BSc in: Engineering Physics, KTH Royal Institute of Technology
Bachelor Thesis: Control of Unmanned Aerial Vehicle (UAV)
Modeling of a quadrotor with the identification of its parameters. Designed the position and the attitude of the quadrotor PID controllers. Thereafter performance and robustness were validated during simulation for the PID controllers. Lastly, PID controller for the attitude was also tested on a quadrotor with an acceptable time response.
Professional History
December 2015 - 2021
Stationsvärd MTR
Got the highest possible score in all my performance surveys on service and sales.
Oversee the station and help customers purchase tickets and find the routes to their destinations.
July 2011 - August 2013
Shop Assistant, Desigual
Built personal relationships with costumers, understood their needs and made clothing recommendations according to them.
Took initiative to assist colleagues when in need and lifted team spirit under stressful situations.
November 2009 - July 2011
Cashier & Kitchen Staff, McDonalds
Received employee of the month.
Best seller.
First contact with customers and took their order.
Checked that all transactions were performed correctly before shift end.
Worked in kitchen, during shortage of staff.
Attitude towards life:
Aut inveniam viam aut faciam
I shall either find a way or make one — Hannibaʿl Baraq
Languages
Swedish
Excellent
English
Excellent