Alisher Kendzhaev - CV DL Engineer

Huawei Moscow, Russia CV Engineer March 2025 – Present – Designed and conducted pose estimation research on RGB-D with ground truth; developed a pipeline from synchronization and retiming to quality metric calculation. Solved the irregular FPS problem: aligned by timestamp, interpolated poses with oscillation regularization. – Developed a driver head position prediction algorithm based on facial key points (eye center), considering occlusions and optical distortion from the display while estimating visible jitter. – Explored and implemented low-pass IMU filtering algorithms with hyperparameter tuning via Optuna and pitch angle prediction for a 150ms horizon. Computer-Aided Design Systems Lab Moscow, Russia DL Engineer February 2024 – February 2025 – Generated synthetic data using FLUX+ControlNet: created weather and optical scenarios (night/snow/rain/mud/glare, partial occlusions) while preserving geometry. – Developed and implemented an auto-annotation system based on CLIP and GroundingDINO to search for and detect cars and license plate regions. – Implemented a Qwen-VL-based system for automatic generation of batch text specifications, improving recall on rare cases by 6-9%. – Designed a detection system for static CCTV using synthetic and real data. CAVISE Moscow, Russia Engineer December 2022 – January 2024 – Integrated and tested MACAD-Gym: implemented a multi-agent RL platform over CARLA for training/evaluating autonomous driving policies. – Optimized V2V/V2X collaborative perception algorithms, improving perception quality and agent action consistency in noisy communication conditions. – Analyzed CARLA maps and optimized training dataset preparation, speeding up preparation by almost 2x.

National Research University “Higher School of Economics” Moscow, Russia Faculty of Applied Mathematics, Mathematics – Bachelor of Mathematics, Applied Mathematics Program.

Pytorch, Lightning, Computer Vision, LLM, ML, OpenCV, Python, Bash, pandas, Docker, FAISS, FastAPI, Transformers, Deep Learning, Latex, Unix, SQL, LoRA, Flux, VLM, ONNX, TensorRT

Английский

An engineer in machine learning and computer vision with experience in AR/VR systems. Specialist in the
development and implementation of algorithms for tracking, recognizing and filtering data, as well as working
with LLM and VLM.