**The problem**

It’s critical to take into account the issue of excess manure produced globally by animals as we work to address the pressing issue of climate catastrophe and adopt regenerative farming techniques. The soil, a non-renewable natural resource, is no longer healthy as a result of human dependence on synthetic fertilizers. We should think about the original fertilizer found in nature instead. It’s important to remember that the bulk of farms—roughly 84%—are modest enterprises with an area of fewer than two hectares. On the other hand, over 50 hectare farms make up just 1% of all farms. Nonetheless, the majority of the greenhouse gas emissions brought on by manure storage and application to the soil are attributable to these larger farms. In actuality, although synthetic fertilizers account for almost 20% of worldwide direct agricultural emissions, manure only accounts for 7% of those emissions. Manure emissions must be decreased through improved management and efficiency in order to counteract this. By promoting the use of manure and organic soil amendments, we can effectively address the issue of excess manure while reducing reliance on synthetic fertilizers and ultimately mitigating greenhouse gas emissions from agriculture. In order to do this, in this Omdena-Nitrolytics project, will be working on this subject using a data-centric approach.

**The project goals**

We aim to develop a tool to evaluate soil water retention and water-holding capacity when manure, charcoal, and compost are utilized. The main goals of this Omdena-Nitrolytics Challenge are:

* Relevant data collection through different possible sources.

* Demonstrate organic fertilizer use and how it improves water retention capacity.

* Give farmers insights into crop water demands in real or near real-time to maximize water conservation.

* Demonstrate enhanced capabilities with biochar application, and support intelligent irrigation.

Determining the Water Retention Capacity of Naturally Fertilized Soil Through Machine Learning

In exploring music recommendation systems, I worked with the Million Song Dataset from The Echo Nest. This hefty dataset consisted of 3 million records from 3337 unique users and 620 songs.

The first step was data transformation using label encoding, which allowed me to make sense of the anonymous user\_id and song\_id. My focus was the 'play\_count' feature, offering insights into user behaviors. Visualization showed most users had a low play count, which suggested either disinterest or a wide music choice.

Among various recommendation algorithms, the user-user similarity-based model outperformed others. Tweaking parameters like cosine similarity threshold led to a boosted F-1 score, even if it increased computation time.

The study provided valuable insights into user behavior in the $25.9 billion global music industry. Notably, I observed trends that helped segment users from casual listeners to audiophiles.

Challenges included the risk of creating a musical echo chamber due to feedback loops in recommendation systems. Possible solutions involved rewarding random music selection or diversifying the training data.

Potential future work could link music choice with the time of day or the listener's task, opening possibilities for predicting user behavior and enhancing digital experiences. Thus, my journey with music recommendation systems was as enlightening as it was challenging.

Music Recommendation System. Capstone Project

As the Lead Designer for the NAVIDOM project, my role involved end-to-end design, from in-depth market analysis to validation of the design post-launch. This project was a comprehensive digital solution aimed to facilitate users' journey of property search and acquisition.

The process started with extensive research, including competitive analysis and first-hand user observations during property searches. These activities provided a deep understanding of user needs and pain points, which drove our design decisions.

I then led the ideation phase, creating a user-friendly and efficient design through sketches and wireframes. This design underwent a series of refinements based on user feedback gathered during the interactive prototyping stage.

Working closely with a UI Designer, we focused on visual design, creating an aesthetically pleasing and intuitive interface across native apps and responsive web versions. We used high-fidelity wireframes to ensure a seamless blend of functionality and visual appeal.

Once the product was launched, we collected user feedback and analyzed user interactions to suggest future enhancements. Throughout this entire process, I collaborated with the development team to ensure high-quality UX and visual design.

The NAVIDOM project for ING Bank Slaski was a prime example of how in-depth user research, collaborative design, and iterative testing can result in a product that effectively meets user needs and business objectives.

NAVIDOM for ING Bank Slaski

AI Product Designer

Part 1. UX Design meets Data Science. Designing AI products with users in Mind.

Unveiling the silent role of UX design in ML Ecosystems.

The era of disappearing interfaces. How AI and Machine Learning are crafting a more human, less visible experience.

Data-Infused Design: A Shinkansen experience hackathon victory and the importance of Exploratory Data Analysis.

MIT. First position in Hackathon. The Hack Linguist

The goal of the problem is to predict whether a passenger was satisfied or not considering his/her
overall experience of traveling on the Shinkansen Bullet Train.

Awards: 2012 Design Management Europe award for public services

Sabina Sujecka

Artificial Intelligence for Impact

Building advanced AI solutions for impact-driven organizations with talented Engineers passionate about making this world a better place.


Skills & Tools Covered:
Leading Research & Knowledge domain 
Streamlit & FIGMA

The project's main theme is to investigate whether assigning personality to a machine or robot adds value, affects task efficiency, and helps build relationships and a unique experience. The conclusions from the presentation will be applied in everyday research and marketing practice. An adapted communication style influences user evaluations of a product or service. In the future, an assistant's personality will play a significant role, becoming part of the brand's DNA, much like familiar solutions such as Cortana, Alexa, or Siri.



Skills & Tools Covered:
Psychological Personality research
MVP
User tests 
Presentation at Orange headquarter

Personality of AI

Embarking on an exploration of music recommendation systems, I analyzed the Million Song Dataset from The Echo Nest. This deep dive took me into the world of user behaviors, algorithmic intricacies, and the challenges and opportunities of personalizing music experiences. From dissecting user play counts to experimenting with recommendation algorithms, and mitigating the risk of creating musical echo chambers, this journey was as enlightening as it was challenging. The insights gleaned hold potential for future advancements in the music industry and enhancing digital user experiences.


Skills & Tools Covered:
Similarity Based Recommendation System
Matrix Factorization Based Recommendation System
Clustering Based Recommendation System
K-Means Clustering
Hyperparameter Tuning

As the Lead Designer on the NAVIDOM project, I managed a process spanning extensive market research to post-launch validation. This endeavor aimed to create a user-friendly digital solution to streamline property searches and purchases. I crafted wireframes, conducted user testing, and coordinated closely with a UI Designer to ensure an aesthetically appealing and intuitive interface. 

This project showcased the value of comprehensive user research, team collaboration, and iterative testing in developing a successful product.

Navidom

FR International publication number PCT/IB2016/001894 · Filed Dec 14, 2016

The invention relates to a method of processing status data captured from an electronic device, to a remote server, the electronic device comprising a button actionable to switch between at least a first state and a second state, said first and second states being associated with respective status tags in a database of the electronic device.

“Processing of status data in an electronic device”

Women in AI

The deceptive machine which maliciously switches off house lights - the impact of character in human-machine communication.

AI Interaction & Product Designer at Innovation Management

Customer Experience and Service Design External Expert

Developed a cost-effective model for international knowledge transfer resulting in 3 new service concepts for public bodies in Silesia

Sabina Sujecka

2023