How AI implementation will influence Thaumatec? – Interview with Michał Zgrzywa, Director of AI @ Thaumatec

Michał Zgrzyw

Michał Zgrzywa, Director of the Artificial Intelligence department is already working in Thaumatec some time, so we couldn’t miss the opportunity to ask him few questions about the AI, it’s influence on the world and our company! Enjoy your reading!

Why can AI be important for the current and future customers of Thaumatec?

Michał Zgrzywa, Directof of Artificial Inteligence @Thaumatec:

– There are countless ways in which Data and AI can bring value to our customers, which is why all the largest analytics companies like Gartner or Forester include them on their list of most impactful technology trends for the upcoming years.

They all seem to agree that in the coming years, the AI revolution will bring severe changes to how we are doing business, communicate, develop ourselves, care for us and people around us and many more. The reason why the expected impact is so large, similarly to all the great technological revolutions of the past like e.g., industrial revolution, is that from now on machines will be able to perform tasks that were so far reserved only for humans. The only difference is that the previous revolutions concerned physical tasks, while the AI one relates also to a subset of cognitive tasks like analytics, recognition, forecasting, spotting trends, spotting anomalies, etc. And these tasks will be able to be performed faster, with more precision and at the fracture of cost.

What is necessary for most of the AI use cases is data. And this is where the current and future customers of Thaumatec are in a strong position – the software embedded in our client’s products and the IoT solutions with their HUBs can provide a lot of data. The data that can be turned into value for our clients and their end users.

Where AI already is bringing the value to product organizations around the world are:

  • improving the product ability to analyze and interpret the environment through the measured signals,
  • extending the product functionality with capabilities like image, sound, or natural text recognition,
  • enhancing the product with possibility to recommend basing on historical patterns, forecasts of the future and environment around,
  • automating the product by allowing it to make autonomous decisions based on data,
  • improving the product ability to recognize unusual behaviors or patterns,
  • understanding of the product usage patterns and issues, which leads to better understanding of the end-users needs, improved product and more satisfied customers,
  • understanding of differences between end-users, which leads to better segmentation and personalization of products and services offerings, to increase revenue and satisfaction,
  • analysis of the products operation patterns that leads to predictive maintenance and lower costs of operation,
  • analysis of the products utilization patterns that leads to recognition of abnormal and potentially fraudulent behaviors, which increases the product reliability and security,
  • finally, the product data may become a new product itself, allowing it to generate completely new revenue streams.

Finally, Thaumatec customers will benefit from having the whole skill set: embedded development, IoT cloud development and AI / data science in one integrated team. We will help our clients to move their products onto a new level of development and gain from the AI revolution instead of being threatened by it.

What new skills will we have as a company?

We will strengthen our company skill set in multiple areas.

First, we will introduce the role of Data Scientist. Such a person needs to combine statistical knowledge, understanding of tools and techniques used in Machine Learning, software development skills (Python, R) with business analyst mindset. The most common technologies and techniques that a Data Scientist knows are:

  • Computer vision: object detection, semantic segmentation, image generation; techniques: various architectures of CNN, GAN, transfer learning, autoencoders, TensorFlow, TensorFlow Lite;
  • Natural Language Processing: speech recognition, NL understanding like text summarization, topic modelling or sentiment analysis; techniques: TFiDF, Word2Vec, BERT, GPT-3 and many more.
  • Predictive modeling: time series forecasting, classification, regression; techniques: ARIMA, regressions, random forests, Xgboost, deep learning, and many more.
  • Optimization: Genetic Algorithms, Bayesian Optimization;
  • Recommendation engines: collaborative, content based, hybrids.
  • Anomalies detection: clustering, dimensionality reduction, isolation forests.
  • Simulations: Monte-Carlo, reinforcement learning.
  • Software development: Python (Pandas, NumPy, Scikit-learn), R;
  • Data Visualization (Matplotlib, Bokeh, Tableau, d3).

The second crucial role we will have is the Data Engineer. Their skills are mostly around retrieving, transforming, cleaning, and storing data. Often Big Data. So, the technologies that are quite common for a Data engineers are:

  • All kind of databases (sql, nosql), data warehouses (cloud, on-premises), data lakes and data transformation tools;
  • Cloud IoT tool stacks: Azure IoT Hub, AWS IoT Core;
  • Big data tool stack: Hadoop, Kafka, HDInsights, Spark, Dask;
  • Software development in general (Python).

Finally, the third role around AI projects is Machine Learning Engineer, whose major responsibility is model operationalization. This person builds the pipelines for model training and model deployment. They are also preparing the test and production environments (often dockerised, located in the cloud). The most common technologies are:

  • ML models training and operationalisation: Azure Machine Learning Studio, Amazon SageMaker;
  • Devops tooling: CI/CD tools, Docker, Kubernetes;
  • Software development in general (Python).

Many of the skills can already be found in our company. But there will also be space for personal development and recruitment.

What kind of projects can we support?

I can envision at least three kinds of projects.

The most exciting projects that we will focus on are development of intelligent products for our customers. Here, to our regular competences of building IoT solutions, we will add the part around training and incorporating the intelligent AI models. This will result in building the AIoT solutions (Artificial Intelligence of Things), that have a huge potential of bringing innovative competitive advantage.

The projects from the second category would aim at existing products that would gain significantly from adding an intelligent component to them. The common scenario in this case is as follows:

  1. we would extend the hardware and firmware to start gathering new data from the product,
  2. we would build the infrastructure that allows storing the data in the cloud,
  3. using the new and already gathered data we would train the intelligent AI models,
  4. we would incorporate the models into the web applications, gateways, or the device itself.

Such extremely complex projects as the above categories require exactly the supplier like Thaumatec – a company that has embedded, IoT cloud and AI skills in the one, well integrated team.

The third category of projects would be more focused on only one part – the AI. In such cases we would cooperate with companies developing their product but lacking the Data & AI competences. We would join by taking care of the AI component, thus helping the customer to achieve their goal.