AI Eng

In today’s world, artificial intelligence is a reliable assistant for programmers, creatives, top managers, students, and more. AI tools help save time, optimize resources, perform routine work, and even generate creative ideas, implementing them at a higher speed. The digital sphere is especially saturated with the latest AI solutions. dev.ua continues the series of materials on the use of AI by IT companies. Today, specialists from SharksCode tell about their experience of interacting with artificial intelligence. We asked about their experience and advice on using the company’s AI tools CMO, STO and Art Director.

The new Claude 3.7 Sonnet is 2 in 1: a classic model with instant responses and a model with extended, step-by-step «reasoning.» That is, they did not release a separate model focused on longer «reasoning» times, like OpenAI with o1 or o3-mini, but added it to the standard model.

Let’s think for a moment about the process of learning to drive. Any of us can learn the basic principles of driving a car by watching a few demonstrations. Once we master concepts like steering, acceleration, and braking, we can apply these skills to any vehicle, from a compact sedan to a large truck.

We continue the series of materials on the features of the use of artificial intelligence in Ukrainian IT companies. Today’s speaker is Oleg Panichev, Expert Research Engineer at Ciklum. He started by developing algorithms for digital processing and analysis of medical signals. While still at university, Oleg worked at his first job at the Ukrainian company Yutas, which specializes in the development of medical equipment. In parallel, the AI ​​specialist was engaged in scientific research, in particular, the analysis of electroencephalograms for predicting epileptic seizures. Over time, the specialist began to notice that machine learning methods were increasingly being used in scientific articles instead of traditional digital signal processing methods. «Previously, event forecasting was based on setting thresholds for certain signal parameters without using cross-validation methods. Realizing the promise of new methods, while working on my dissertation, I began to study machine learning and participate in competitions on the Kaggle platform,» he told dev.ua. Oleg has been working with AI for over eleven years now and has experience solving a wide variety of problems, from working with tabular data to deep learning for computer vision and GenAI. He talked about what he does at Ciklum and how the company integrates AI tools into its projects. The specialist also gives practical advice to those who plan to study and work with AI.