From our cutting-edge research and engineering initiatives we develop generalised AI solutions that solve complex tasks.
We pride ourselves on making our solutions easy to use, scalable, and reliable. Everything can be managed entirely by us, or deployed to your environment exactly where it is needed, be it on embedded devices, mobile, or in the cloud ensuring that everything is only an API or function call away.
Use our API for any multimodal search task. By combining various types of data we are able to create tailored and scalable search engines whether you want to search by similarity or semantically in text, video, images or audio, for products, documents, or podcasts.
Efficiently summarise any data to remove the noise and extract only the insights that you need. Our summarization services extract topics and insights from any data source, to give you everything you need at just a glance.
Continuously adapt to your user's ever changing preferences and serve them only the most relevant recommendations. Our dynamic system can start off with only a few data points, ensuring that your product can get traction quickly.
Using reinforcement learning, we are able to automate and optimise decision making in realtime. From control loops to selecting the best set of actions to take in a scenario, our agents find the perfect balance between exploring new strategies, while exploiting what is already know.
Real world machine learning is never the same as a course you took or a video you watched. With the Llama programme, we aim to give you the experience to take you from a newcomer to an absolute badass machine learning llama by working on real projects with us.
Depending on your experience level, we will develop a custom training plan for your first 2 weeks with us. During this time, you will be introduced to your mentor who will be working with you throughout your time with us. After this training session, you will be given your own project which you will take control off, with real world deliverables. Depending on your interests, these projects are either research driven or engineer driven. After about 2 months, you and your cohort will present your work at the LLama ML day.
You must have had exposure to machine/deep learning, but you do NOT have to be an expert at all! Along with this you should not be afraid to program and you need to have passion and resilience. Ideal candidates are post-graduate students or undergraduates in their 3rd year at least. If you are not at university, don’t worry. If you are on the level of 3rd year undergraduate, you can still apply.
Firstly, there was a project titled Zero shot image classification enhancement via text label improvement. This project created a model that can optimize the list of labels given to a zero-shot image classification model, in this case Open AI’s Clip model. The labels are optimized in such a way that the representation of the label is closer in Euclidian space to the target image representation. Another project titled Semantic Summaries from Sequential Media created a system that can detect when the “topic” of sequential data, such as podcasts, has changed based on the trained representations.
Click apply, wait for the call, answer some questions and badaboom you are in.