Position: Computer Vision Research Engineer
Company: Victori-Technology
City: Astana
Employment: offline
Salary: from 800,000 KZT.
Vacancy Description:
Responsibilities:
- Development and implementation of ML models for video and audio data analysis;
- Detection and classification of emotions and psychophysiological states (stress, deception, fear, anger, nervousness) based on facial expressions, micro-movements, gestures, voice timbre, and intonation;
- Building pipelines for pre-processing video/audio, feature extraction (landmarks, audio embeddings, spectrograms);
- Development and implementation of models using CNN, RNN, Transformer, Audio/Video Fusion, LLM/RAG architectures for result interpretation;
- Development of model validation methodologies, including audio-video synchronization, cross-validation, metric evaluation (accuracy, precision, recall, F1-score);
- Experiments with various neural network architectures (Vision Transformers, Audio Transformers, Bi-LSTM/GRU);
- Researching model interpretability and explainability methods for legally significant conclusions (Explainable AI);
- Conducting code reviews of existing ML models and evaluating implementation quality;
- Optimizing, improving, and modifying model architectures to enhance accuracy, stability, and performance;
- Supporting models in the production environment, monitoring results, and fixing bugs;
- Developing APIs and mechanisms for transferring interpreted model results to existing systems.
Requirements:
- 3+ years of experience in ML/CV/NLP/RAG, preferably in video/audio analysis projects;
- Deep understanding of computer vision, audio processing, and multimodal data;
- Experience with PyTorch/TensorFlow;
- Experience in developing and implementing CNN, RNN, Transformer, Vision/Audio Transformers, multimodal fusion;
- Practical knowledge of data preparation, augmentation, and feature extraction (landmarks, spectrograms, embeddings);
- Experience with RAG architecture or LLMs for interpreting ML model results;
- Experience integrating ML models via API with external systems;
- Experience with ML pipelines, MLOps, Docker.
Conditions:
- Office work in the Presidential Park area, no remote work;
- Projects in LegalTech and digitalization;
- Real tasks, quick solutions, and space for initiative;
- Full-time employment, 3-month probation period;
- Schedule: 5/2, 9:00–18:00.
Contacts:
Telegram recruiter contact
