“Our people make us successful. When all of these talented people work together, we are able to reach our goals.”
Alexander Suslov is a Deep Learning R&D Engineer in Nizhny Novgorod, Russia. As someone with a background in IT and tech startups, he wasn’t sure about moving to a global company like Intel—but he says the transition was easier than expected. He is constantly impressed by the level of expertise on his team and feels excited to be working on such state-of-the-art innovations. We had a chat with Alexander to learn more about his experience as part of the OpenVINO™ group in Russia.
What exactly does your group do?
Our group is developing the OpenVINO™ toolkit, which allows developers to deploy deep learning models on Intel® platforms. Specifically, our team is working on development of tools for compression and optimization of deep learning models. These tools allow developers to apply techniques such as quantization and pruning to deep learning models to reduce execution latency and model size. We develop PyTorch-based Neural Network Compression Framework (NNCF) for optimizing deep learning models during training and Post-training Optimization Toolkit (POT) for optimizing pre-trained models without re-training when the developer cannot fine-tune the model. Both tools are part of the OpenVINO™ toolkit. My responsibility is to develop the TensorFlow-based Neural Network Compression Framework for enhanced OpenVINO™ inference.
What major challenges do you face?
Our main challenge is to find a balance between performance and accuracy. Customers are always interested in speed, but when you make something faster, you often sacrifice a bit of accuracy. We are constantly working to achieve optimum results. We want our product to perform quickly, but not at the expense of accuracy.
How do you think your work is impacting the future?
Our group is developing open source products, which means we provide developers all over the world with instruments and tools to build their own solutions. This is quickly growing a strong deep learning developer community around Intel. The brightest minds in every part of the world are using our framework. They are easily able to achieve the best performance on Intel hardware and accurately estimate hardware requirements for their solutions. Our work is helping to improve technological development in many sectors. I believe that as we see decreases in time-to-market, we will also begin to see more Artificial Intelligence applications, which will ultimately transform everyday life.
What makes your team successful?
Our people make us successful. We are a strong, high-performance team with excellent skills in computing and deep learning. When all of these talented people work together, we are able to reach our goals.
What excites you about being part of Intel?
Firstly, I’m working on state-of-the-art algorithms in deep learning. This is exciting! Secondly, I work with wonderful people who challenge me to grow and improve my skills each and every day.
What opportunities do you have at Intel?
At Intel, I have been given the support and encouragement to grow, both personally and professionally. In my group, we are often given the opportunity to attend technical seminars to learn from other developers and leaders in the industry. As a company, Intel is very interested in each employee’s continued professional development. I have seen many people on my own team progress. We are given the ability to change things not just within Intel, but for the entire technological community.
Did anything surprise you when you first joined Intel?
I was surprised by the amount of expertise inside the group. Also, Intel culture is very different from other tech companies. It’s a large global company, so when you first join, you have to learn all the processes that are unique to Intel. I have been impressed with how well these protocols operate to keep communication and workflow on track. It’s very helpful.
What do you appreciate about working at Intel in Russia?
Intel Russia is one of the biggest IT companies in Nizhny Novgorod. Many talented people come here just to work with us, which means we have a very high level of expertise in deep learning. People come to us from Moscow, for example. It may seem strange, because salaries are often higher in Moscow, but people move to this smaller city just to learn with Intel.
What skills do you need to work with Intel?
First, you need a very good mathematical grounding. I also think it’s important to be self-motivated, with a desire to learn and good communication skills. We like people who know about deep learning, but experience in this area is not essential. As long as you are interested and motivated, Intel is willing to teach you on the job.