Ugurkan Ates

Research Software Engineer

Liege Belgium - Istanbul,Turkey(National) +90-053*-***-**** adk8em@r.postjobfree.com

ugurkanates.github.io github.com/ugurkanates

An adaptable and responsible Computer Engineer . Seeking an entry to mid level research & software engineer position in Deep Learning and AI as well as embedded software development which will utilize the organisational and research skills developed through my involvement with the field.

2020 March - 2021 March

2019 January - 2020 March

2019 January - 2020 March

2018 September - 2019 September

2018 June - 2019 January

2018 Jan - 2018 June

2017 September - 2018 June

Research Software Engineer

University of Liege, Belgium

Worked on industrial vision based robots (ABB) . My main contribution was applying vision based deep reinforcement learning experience to complete grasping operation.We developed a system trained exclusively on simulation and adapted system to real world grasping robots with various grippers. Our system used moving camera design with depth time of flight camera.Worked with batch-offline RL(CQL,Batch DQN) and off-policy adaptation techniques to achieve state of art performance with KUKA/ABB FlexPicker robots.

Main gains are: Deep Learning, Reinforcement Learning, Virtual Cameras, Closed loop Control, Robotics,Computer Vision Machine Learning Engineer

Accenture - Industry X.0

I work on Machine Learning & Deep Learning projects as Jr Machine Learning engineer. Worked on real time lane detection with different methods for best inference on embedded devices such as Raspberry Pi 4,Jetson TX2 and Xavier AG. Worked to make openCV python with CUDA modules for best efficiency . Also trying to optimize LaneNet models to see visibility on embedded devices.

Main gains are : NVIDIA Jetson Systems, Object Detection & Lane Detection, Multi Image Stitching, Deep Sort & Tracking Embedded Software Engineer

Accenture - Industry X.0

I worked as embedded software developer under IX.0 project with universal and diverse team.. Working with different kind of clients from industrialized factories to vehicle companies. We provided embedded software solutions. Main expertise I contributed are Yocto Linux project, embedded software with C++17 to different MCU/MPU's, bare metal support if needed, docker and containers support for production level products.

Main gains are : Embedded Software Solutions,C++ and YOCTO Linux, Bare Metal Coding,STM32,FreeRTOS Embedded Sofware Engineer

GTU Hazine Autonomous Electric Car

I worked on our university first fully autonomous vehicle with team of 30 student from different disciplines. As part of official Turkish autonomous vehicle contest held in TEKNOFEST 2019, I was part of LIDAR and simulation teams. With Velodyne-16 we have implemented and tried different localization algorithms for race track. With help of ROS and sensor fusion from our computer vision team we built a motion planning algorithm to race the track correctly. Also worked with Unreal E4 and CARLA/RoadRunner to built exact replica of track and do autonomous test rides on it before implementing on actual car. Main gains are : LIDAR, ROS-Robot operating system, CARLA - Autonomous Vehicle Simulations,Sensor fusion with Computer Vision & Embedded Solution & LIDAR.

Software Engineer Internship

Turkish Aerospace Industries ( TAI - TUSAS)

I worked as embedded software engineer at TAI for 6 months as part of engineer internship programme.I have developed Bus Monitor for ARINC communication system. Worked closely with telecommunication protocols such as I2C,UART,SPI,RS422 and RS485. Implemented GUI interface for monitoring all communication and implemented socket communication with C++ also developed a simulation to emulate USB hub interaction within real commercial aircrafts. Main gains are : Communication protocols(I2C,UART,RS485,SPI),MCU/MPU pprogramming, interacting with wide diverse sensors. Part Time Software Developer

Softtech AS

I worked at SoftTech one of biggest Turkey based software developing house for 5.5 months as Software Developer. At SoftTech I have worked on multiple different projects. Worked on PEGA Marketing & Customer Relation Management departments for 2.5 month. I learned lot about Java backend services. Also worked with Python on Customer Prediction algorithms and software about company services used inside offices of IsBank. Main gains are : Software methodologies in action(AGILE,Scrum), Java and GIT interactions, working with JIRA. Software Developer of CanSat 2018 Team

AAS CanSat Competition - AEROLITE Team, Austin Texas Creating Ground Control Software of American AeroSpace Society & NASA CanSat 2018 AeroLite for Gebze Technical University. Using Java FX for graphs and JPython for data transactions. Creating and maintaining sustainable code for futures. All created so far helps open-source community.

Also helped Project Management and Maintaining.

All code and documentation about this project resides on GitHub page. English – Native level speaking,communication and

professional working proficiency

Turkish– Native language

Experience with C,C++ and modern improvements of

C++ (11-14-17) and Python 3

Experience with Object Oriented software methods

and design patterns

Deep understanding of Machine Learning and Deep

Learning topics

Embedded programming & sensor fusion on

autonomous systems

Research area is Deep RL, worked on path planning and robotic manipulation

Worked with Embedded Linux,NVIDIA Jetson and

CUDA systems and various ARM systems over 2 years. Have basic knowledge with gaming engines such as

Unity and Unreal, contributed to Unity ML Agents

library

Familiar with OpenCV,Tensorflow,PyTorch

Proficient with various LINUX distros,Docker and

version control systems.

Summary

Work Experience

Languages

Technological Proficiencies

Created with

- 10 July 2020

2019 August - 2019 December

2019 January - 2019 May

2017 - 2018

Long-Term Planning with Deep Reinforcement Learning on Autonomous Drives Research Paper

Arvix Link for paper :(https://arxiv.org/abs/2007.05694) https://ieeexplore.ieee.org/document/9259811 - Published by IEEE Journal My single author AI paper got accepted to one conference and one journal in 2020. Submitted to Association for the Advancement of Artificial Intelligence 2020 Fall Symposium(AAAI 2020) Washington,DC November 11-14,2020

https://sites.google.com/view/longtermplanrl

Competitive Motion Planning with Deep RL

Graduation Project 2

As part of my final project 2 I have developed a Deep RL system using model free policy gradient methods(TRPO,PPO,DDPG) . I used AirSim which is a very complex drone/car RL environment with continuous space. I participated in Microsoft NeurIPS 2019 challenge on planning an autonomous drive control system and doing competitive planning. combines Deep RL with Computer Vision.Using forward and downward camera I take frames and after some cleaning up process images feed intoCNN. I have experimented DDPG and PPO and was able to get some agents behave better than minimum jerk trajectory. Main gains are : Convolutional Neural Network, Deep Learning, Reinforcement Learning, Double Q Learning with Replay Memory Experience,Deep Deterministic Policy Gradient ability to implement latest papers, being able to track and train on remote machines, CUDA and PyTorch, Using docker to create ready RL products. Smart Health Band with Machine Learning(Human Activity Recognizer) University Project *, Istanbul Turkey

We developed a fully functional smart wrist health band embedded device that interacts with native android app. With this device users can able to track their movements on native app on android or web browser. Can track daily movements,set alarms for walks,detect if you idle too long, track your route via GPS, do ML computations to keep track of your movements classification, alarm if your heartbeat is too low/high to SMS/call your friends or ambulance. With ESP32 as MCU and tracking sensors we created fully functional battery chargeable device that could be sold to public market. For more information please check https://bilekpartner.wixsite.com/website

Main gains are : MCU programming with C++, Interrupts and parallel threaded programming parallel sensor communications, native android app with Java and live graphs simultaneously, web services to demonstrate logs/reports, ML algorithms (linear classification)

High Power Rocketry Design & Software Competition for TeknoFest ISTANBUL 2018 University Project *, Istanbul Turkey

I worked as Software developer with our team competing for Rocket Competition held by Teknofest Istanbul and Turkish Technology Team. Worked on probabilities launch of rocket, fall tests,flying tests and more. Used other open sourced software as well and it got good results .We flew rocket (1500M ~ 1.5KM) and used parachute to save rocket. All code related to it open sourced and properly documented on my personal GitHub page. We were finalist of this competition. Main gains are : C++ programming wth MCU, live graphics with Python, device drivers & OpenCV for real time video display. CanSat NASA - TURKSAT Mini Satellite Competition

University Project * Austin - Texas, USA

I have worked on Ground Control System Design,Communication and Data Handling Subsystem Design and Management parts of object.My primary goal on this object was that getting all data packages while CanSat module flying and communicating with Ground Center(our base laptop). Also developing Ground Control Software and Interface with Java using Object Oriented Methods. Competition is helded by NASA in June 2018 all code related to is on my GITHUB account. Main gains are : Python programming, C Programming, FSA Design & Implementation, Java programming, MCU and Linux knowledge

2020 - 2021 Jan

2015 - 2019

Master MSc in Computer Science

University of Liege, Belgium

I took following courses Machine Learning Introduction, Artificial Intelligence and Search Algorithms, AI in law, Control Systems as I was guest student for half semester as part of research programme that I was working full time. ( One Semester ) Bachelor Degree in Computer Engineering

Gebze Technical University - Turkey

Graduated in June 2019.

Took very important CSE related courses : Embedded Design and Finite State machines(Logisim,Verilog,MCUs), Object Oriented Programming, Design Patterns, System Programming with Linux, Software Engineering and Advanced Methodologies, Digital Image Processing, Algorithms and Data Structures, Performance of Algorithms and Dynamic Programming, Programming Languages(LISP, Q#,Prolog), Deep Learning & Machine Learning Projects/Courses/Certificates

Education

Contact this candidate