Embedded AI & Control Systems Engineer focused on:
- computer vision
- robotics
- real-time systems
- experimental control engineering
- edge AI inference
- distributed intelligent systems
I enjoy building complete systems — from mathematical modeling and control design to embedded firmware, AI inference pipelines and cloud-native backend infrastructure.
Currently pursuing a B.Sc. in Computer Science at the National University of Science and Technology POLITEHNICA Bucharest, Faculty of Automatic Control and Computer Science.
- Embedded AI
- Computer Vision
- Robotics & Autonomous Systems
- Optimal & Adaptive Control
- Real-Time Systems
- System Identification
- Edge Inference
- Retrieval-Augmented LLM Systems
- Intelligent Monitoring Systems
Real-time multimodal exam monitoring platform combining:
- gaze estimation using MediaPipe Face Mesh (468 landmarks)
- adaptive Kalman filtering
- dual YOLOv8 pipelines for smartphone and smartwatch detection
- audio activity detection + Romanian speech transcription
- motorized gimbal classroom scanning
- multi-student orchestration and event persistence
- 25+ FPS CPU-only inference
- <820 MB RAM usage
- 92.4% gaze-direction accuracy
- 88.6% smartphone detection accuracy
- 92.5% smartwatch detection accuracy
- fully local GDPR-oriented processing
First-author paper published at RoEduNet 2025 (ISI Indexed Conference).
Repository:
- private research repository
- extended into 70+ page Bachelor's thesis
Repository:
3-DOF active beverage stabilization platform designed for experimental comparison of classical, optimal and adaptive control strategies.
- dual-MCU architecture
- Teensy 4.1 + Arduino UNO
- BLDC FOC motor control via Moteus drivers
- FDCAN communication
- N4SID system identification
- reduced-order modeling
- disturbance rejection benchmarking
- experimental controller comparison on real hardware
- PID
- RST
- LQG
- MRAC
- automated testing framework
- custom Composite Performance Assessment (CPA) metric
- MATLAB identification + simulation pipeline
- embedded real-time implementation
Paper currently under preparation.
Repository:
Research project exploring retrieval-augmented prompting for ordinal NLP classification using local open-weight LLMs.
- class-balanced semantic retrieval
- frozen sentence embeddings
- retrieval-augmented prompting
- semantic nearest-neighbor search
- comparative evaluation across open-weight LLMs
- Mistral
- Gemma
- Qwen
Achieved ~90% classification accuracy without model fine-tuning.
Paper under preparation.
May 2025 – April 2025
Worked on applied research projects involving:
- AI-based anti-plagiarism systems
- ESG analytics using LLMs
- retrieval-augmented document intelligence
- cloud-native AI pipelines
- distributed backend services
May 2026 – Present
May 2025 – April 2026
Worked on applied research projects involving:
- AI-based anti-plagiarism systems
- ESG analytics using LLMs
- retrieval-augmented document intelligence
- cloud-native AI pipelines
- distributed backend services
- FastAPI
- Docker
- AWS EC2
- PostgreSQL
- Python
- LLM APIs
- embeddings & retrieval systems
RoEduNet Conference 2025 — ISI Indexed First Author
- 🥇 1st Place — Scientific Communications Session (SCSS 2025)
- 🥇 1st Place — TechChallenge & RoboChallenge 2025 (Freestyle) & RoboTEC (Freestyle)
- 📄 First Author — RoEduNet 2025 (ISI Indexed)
- 🔧 Open Source Contributor — Rust
rencfs
| Domain | Technologies |
|---|---|
| Languages | Python, C, C++, Java, TypeScript |
| AI / CV | YOLOv8, OpenCV, MediaPipe, PyTorch, LLMs |
| Embedded | STM32, Teensy 4.1, Arduino, FDCAN |
| Control | PID, RST, LQG, MRAC, Kalman Filtering |
| Backend | FastAPI, Docker, PostgreSQL |
| Cloud | AWS EC2, Linux, OpenShift |
| Frontend | React |
| Tooling | Git, MATLAB, Simulink |
