Actionable Insights with Industrial-Grade Video Understanding
AI for the Physical World. At Ramblr, we go beyond superficial video analysis to extract deep context from egocentric videos. Our technology provides a comprehensive understanding of actions, individual objects, and their relationships. Prompt Ramblr’s AI assistant to unlock precise insights and pinpoint specific moments in thousands of hours of multimodal videos captured from a first-person perspective. Are you excited to become a Ramblr and join us at the intersection of AI and spatial computing? If so, you can apply directly to the job posting or use the open application form.
We look forward to hearing from you !We are looking for a strong deep learning scientist who can develop and monitor state-ot-the-art AI models. You quickly and efficiently probe and evaluate deep learning models for new use cases and help bring them into production. You quantitatively evaluate the performance of models and optimize the interplay between different components of our deep learning stack.
You work closely at the intersection of research and product development. You are a key player in extending our offering for internal use and customer use-cases - field testing and iterating quickly.
Passion for solving the hard problems in deep learning
Required:
Data driven QA: definition of useful metrics, implementation of processes to measure metrics, data visualization
Machine Learning: model analysis, data loading pipelines, data monitoring, neural network architectures (especially Transformer / CNNs), model training, PyTorch
Computer Vision: CNNs, Vision Transformers, spatio-temporal data, image- and video embeddings, image augmentation
Experience with scientific python libraries such as: numpy, openCV, matplotlib, scipy, scikit-learn, scikit-optimize, pandas, seaborn
General: git VCS, code reviews, development on Linux, distributed computing concepts
Proficiency in python-based collaborative software engineering: follow consistent style-guide, clean design patterns, self-documented code, unit/integration tests, type annotations
Optional:
Natural Language Processing: Transformer architectures, vision-language alignment, prompt engineering
Video understanding: Action understanding, activity prediction, scene graph prediction
