Computer Vision Engineer

Key Responsibilities

• Core Video Analysis:

   Implement and optimize fundamental video processing algorithms, including Optical Flow, Motion Estimation, and Background Subtraction to handle dynamic sports footage.

• Spatio-Temporal Modeling:

   Develop models that understand time as well as space, utilizing 3D CNNs, RNNs/LSTMs, or Temporal Transformers to recognize actions and events over sequences of frames.

• Object Tracking & Re-Identification:

   Build robust tracking pipelines using industry-standard algorithms (e.g., Kalman Filters, SORT, DeepSORT) to maintain identity of players and objects across occlusions.

• Advanced Architectures:

   Research and integrate state-of-the-art models, including Vision Transformers (ViT) and Attention Mechanisms, to improve accuracy beyond traditional CNN limits.

• Agentic AI Workflows:

   Assist in designing Agentic AI systems where autonomous agents plan multi-step video analysis tasks (e.g., deciding when to focus on specific game events) with minimal human intervention.

• Data & Pipeline Strategy:

   Manage video datasets and collaborate with the engineering team to deploy efficient inference pipelines.

Required Skills & Qualifications

• Education:

   Bachelor’s or Master’s degree in Computer Science, AI, Data Science, or a related field.

• Core Computer Vision:

   Strong understanding of traditional CV concepts:
• Image Geometry & Camera Calibration
• Feature Extraction (SIFT, SURF, ORB)
• Image Filtering & Edge Detection

• Deep Learning for Video:

   In-depth knowledge of neural network architectures tailored for video:
• CNNs (ResNet, EfficientNet) for spatial features.
• Sequence Models (RNN, LSTM, GRU) for temporal dependencies.
• 3D CNNs (C3D, I3D, X3D) for spatiotemporal feature learning.

• Transformers & Attention:

   Understanding of Self-Attention mechanisms, Vision Transformers (ViT), and how they differ from convolutional approaches.

• Programming:

   Proficiency in Python with libraries like OpenCV, NumPy, Pandas, and Scikit-learn.

• Frameworks:

   Hands-on experience with PyTorch or TensorFlow.

Good to Have (Bonus)

• Experience with Agentic AI frameworks (e.g., LangChain) applied to visual tasks.
• Knowledge of Multimodal AI (Video + Audio/Text).
• Familiarity with model optimisation tools (TensorRT, ONNX) for real-time video inference.

Application Link: