Adrian Dumitru Ciotec

• Machine Learning (Deep Learning, Reinforcement Learning) for car signals: jerk, electrical signals
• Submitted 3 valid patent ideas
• Submitted an article to a journal: https://arxiv.org/abs/2501.17653
• Docker
• Azure ML cloud
• LLMs & prompt engineering
• OpenAI
• Focus on prototype work (MVP, PoC)
• Research and backend software development
• Scrum Agile
• Courses on Coursera on ML topics (including agentic AI)

Machine Learning for Computer Vision - autonomous driving platform:

• Writing yaml pipelines and Python code with Tensorflow 1.15 and 2.3
• Neural-Network architecture improvements
• Running evaluation and training experiments
• Data-topics (ex. finding errors in image labels)
• Participating in interviews with machine-learning topics
• Coaching students on machine-learning topics

• Mentoring Udacity's students for Deep Learning nano-degree in my free time (independent contractor)

• Working with Python to build speech-to-text (STT) feature - research and decision about STT model and engine to be used:
• Mozilla Deepspeech (Tensorflow)
• DeepSpeech 2 (PyTorch)
• Wav2letter (Torch)research and decision about hardware selection
• Training neural networks for speech-to-text
• Collaboration with a mathematician and implementing of scientific algorithms in Matlab, Python, Julia
• Creating and configuring lxc/lxd containers to run specific deep-learning frameworks
• Worked with Docker
• Worked mostly under Linux

• Embedded C developer for Lenovo ethernet switches
• Worked with: C; Linux OS; Bash; GDB debugging; Python;
• Source control: IBM ClearCase
• Working in a waterfall setting
• Daily tasks: writing code for new features, correcting bugs, fixing static code analysis issues (continuous integration)

• Worked with C, Python, Git, Bash, Linux, Android ADB;
• Understanding and extracting power management algorithms
  from Android kernel;
• Instrumenting Android kernels for power measurement by changing Android drivers;
• Evaluated Machine Learning algorithms using Python and TensorFlow;
• Submitted an article to an internal Intel Software Conference (first author) which was accepted for presentation;
• Worked as a RINF subcontractor at Intel Romania for the first 10 months and was offered a full position inside company afterwards;

• Teaching Digital Signal Processing (DSP) course and laboratory
• Teaching Pattern Recognition laboratory
• Teaching Decision and Estimation in Information Processing (similar to statistical signal processing) laboratory
• Author and coauthor of over 15 scientific papers in the fields: artificial intelligence, computer vision, remote sensing, time-series
• Google Scholar: https://scholar.google.ro/citations?user=HJhhKHsAAAAJ&hl=en
• Laboratories made extensive use of Matlab

• Research, design and develop a software application (PCC) to control Phantom high speed video cameras and manipulate .cine video files
• Application was developed in C#.NET and made use of Phantom SDK
• It also includes measurement tool with auto-tracking functionality, firmware updater;
• The main application (PCC) is used by worldwide costumers;
• Worked with:

- Image processing

- Windows Forms

- C# Marshalling

- Mixed C# / Native C debug

- Matlab

- Source control: TFS;

• Redesigned part of PCC application;
• Designed and implemented from scratch auto-tracking functionality and change detection features;
• Daily tasks:

-Implementing new features;

-Participating at Phantom SDK design;

-Taking design decisions for .NET application features;

-Fixing bugs;

-Writing Phantom SDK demos and small tools in Matlab and C++;

• Testing applications;
• Developing .NET applications using C# and Windows Forms mostly;
• Source control: SVN;
• Note: part-time job during faculty years;