Computer Vision Engineer / Machine Learning Engineer

Описание

Mikael Yeghiazaryan myeghiaz@illinois.edu | LinkedIn | GitHub Research Assistant at University of Illinois Urbana-Champaign Education University of Oxford Oxford, UK Bachelor & Master of Engineering Science October 2018 – June 2022 •Graduated with a 1 st Class Honours; ranked 15 out of 159 • Received the Swire Scholarship for excellent academic performance “Quantum” Gymnasium Yerevan, Armenia International Baccalaureate Diploma (High School) September 2016 – July 2018 •Overall IB score: 41/45 Research Projects Safe Autonomous Unmanned Aerial System |UIUC July 2024 – present • Supervision: Dr. Naira Hovakimyan. • Description: Our work focuses on implementing an autonomous unmanned aerial system that is capable of safe operation in scenarios where a serious system fault occurs, e.g. GPS signal is lost or an actuator is stuck. My responsibility is to develop a simulator for air taxis in cluttered urban environments using CARLA. • Skills: Simulation (CARLA, Gazebo), ROS. Augmenting Aerial Imagery using Vision-Language Models |CMU September 2023 – present • Supervision: Dr. Fernando De la Torre and Dr. Jessica Hodgins. • Description: Our goal is to enhance aerial/satellite imagery employed as training data for vehicle detection. Our primary focus revolves around leveraging advanced text-to-image models, such as Stable Diffusion, to generate and augment data in a controllable manner. We are also running adversarial attacks on these generative models to assess their stability, and explain detector performance using language. • Skills: diffusion models, generative AI, vision-language models, large language models, prompt engineering. Practical Adversarial Attacks on Aerial Imagery |CMU September 2022 – December 2023 • Supervision: Dr. Fernando De la Torre and Dr. Jessica Hodgins. • Description: This pro ject focused on devising practical adversarial attacks against aerial vehicle detectors. • Skills: adversarial attacks, 3D differentiable rendering (PyTorch3D, Kaolin), synthetic data generation, Deep Neural Networks for ob ject detection, Blender, teamwork, presentation skills. 4 th year Master Thesis Pro ject |University of Oxford September 2021 – June 2022 • Title: Learning generalizable keypoints for ob ject pose estimation. • Supervision: Dr. Joao Henriques and Dr. Dylan Campbell (Visual Geometry Group, University of Oxford). • Description: I worked on creating a novel approach for unseen ob ject pose estimation using general features and low-resolution CAD models of unseen ob jects. • Skills: differentiable rendering (PyTorch3D), 3D geometry, work ethics. 3 rd year Group Pro ject |University of Oxford September 2020 – May 2021 • Title: Formula Student Electric Vehicle (EV) Race Car Design. • Supervision: Dr. Dan Rogers (University of Oxford). • Description: I worked on the design of the electronics and control systems for a Formula Student EV race car. MIT GTL Pro ject June 2016 – August 2016 • Title: Movie Feedback Classification Using the Perceptron Algorithm. • Supervision: Hayk Saribekyan (MIT student at the time of the pro ject). • Description: I trained a perceptron classifier which classified movie reviews (given as a raw text input) either positive or negative, and managed to achieve 96% accuracy on the test dataset. Work Experience Research Assistant Urbana-Champaign, IL, USA Advanced Controls Research Laboratory, University of Il linois Urbana-Champaign July 2024 – present •Supervised by Dr. Naira Hovakimyan. Research Assistant Pittsburgh, PA, USA Robotics Institute, Carnegie Mel lon University September 2022 – July 2024 •Supervised by Dr. Fernando De La Torre and Dr. Jessica Hodgins. Research Assistant Oxford, UK Engineering Department, University of Oxford November 2021 — January 2022 •Supervised by Dr. Min Chen (University of Oxford). • Assembled and annotated a dataset of 3D images. Research Intern Oxford, UK Oxford Robotics Institute, University of Oxford June 2021 — August 2021 •Supervised by Dr. Maurice Fallon (ORI). • Integrated a 3D semantic segmentation model into a quadruped robot as a ROS (robot operating system) node. Honours and Awards 1 st Class Honours Oxford, UK University of Oxford October 2018 – June 2022 •I achieved and graduated with 1 st Class Honours (equivalent to 4.0 GPA) in all years of examinations at the University of Oxford. • Awarded the Swire Scholarship for exceptional academic performance at my college (University College Oxford). Honorable mention in The International Physics Olympiad Jogjakarta, Indonesia International Physics Olympiad, 2017 July 2017 Silver medal in The International Zhautykov Olympiad (Physics) Almaty, Kazakhstan International Zhautykov Olympiad, 2017 January 2017 Community & Leadership Secretary and President of the Oxford University Armenian Society Oxford, UK Oxford University Armenian Society (OUAS) October 2020 – June 2022 •I set up events for the society members. • I engaged in the promotion of the society’s events at the university level. Lecturer of Engineering Science Yerevan, Armenia “Quantum” Gymnasium August 2019 • I lectured engineering of electronics at my alma mater’s summer camp for high school students. President of the Experimental Physics Club Yerevan, Armenia “Quantum” Gymnasium October 2016 – March 2018 •I founded a club for conducting experiments in physics and helped other students with their assignments. Skills Programming Languages : Python, C/C++, Matlab. Machine Learning & AI : Pytorch, Natural Language Processing, CLIP, Diffusion Models, Tensorflow, Keras. 3D Computer Vision & Graphics : PyTorch3D, Blender, CARLA, PyBullet, Kaolin. Operating Systems : skillful in Linux (Ubuntu 20.04). Other : Git, Prompt Engineering, ROS, Arduino, Raspberry Pi 4, Flask, AWS, Anaconda, L A T EX, PostgreSQL. Research : critical thinking, teamwork, communication and presentation skills, work ethics. Miscellaneous Languages : Armenian (native), Russian (native), Belarusian (native), English (fluent), Spanish (beginner), German (beginner), Arabic (beginner). Music : Playing the violin since age of 6, laureate of international contests, invited performer at Vladimir Spivakov’s festival at age 10. Chess : Achieved Class A Elo ranking at age 12, now playing recreationally (Lichess ID Michael Yeghiazaryan).