Principles of ML & Deep Learning
TensorFlow models, Computer Vision, CNN, NLP
Hello, I'm
Yosriko Rahmat
AI and Data Enthusiast
Hello, I'm
AI and Data Enthusiast
Get To Know More
6 Month
AI/ML Engineer
Informatics Student
Bangkit Academy 2023
Passionate AI and Data enthusiast with substantial hands-on experience in AI projects, model implementation, and machine learning applications. Currently active as a researcher and lecturer assistant at my University, where I contribute my expertise and passion towards advancing knowledge in these dynamic fields. My professional journey includes working on diverse AI projects such as developing predictive models, natural language processing, image recognition systems, and data-driven decision-making tools. Known for my adaptability and strong learning abilities, always eager to embrace new challenges that push me to enhance my skills.
Explore My
TensorFlow models, Computer Vision, CNN, NLP
Build, train neural networks, prevent overfitting
CNNs with Conv2D, pooling layers,improve neural network, image augmentation, ImageDataGenerator, labels based on the directory structure
Text categorization, word embeddings, LSTM, RNNs
RNNs, CNNs for time series, sequence forecasting
Browse My Recent
Role: Scrum Master & AI Engineer
Tools: TensorFlow. Kotlin, Vue.js
Description: Develop a student attendant system that uses QR Code, Geo Location, and Face Authentication.
Result: Able to detect blurriness, detect and recognize faces.
Role: AI Engineer
Tools: Pandas, TensorFlow, Matplotlib
Description: Machine learning-based Android application using image recognition to scan Cultural attribute.
This application is also expected to be educational material for local and foreign travelers about Balinese masks.
Result: Able to classify Bali mask with more than 98% accuracy.
Role: Code in Arduino IDE, Connect MQTT
Tools: ESP2866, Node-RED, PH Meter, TDS, WaterFlow, and other support tools
Description:
On-going IOT project that using five sensors to monitor plants in the garden enables accurate actions to improve plant health. The sensors include the DHT11, pH meter, TDS sensor, water flow sensor, and ultrasonic sensor. For visualization, a Node-RED dashboard is used,
which also allows users to control the system with interactive buttons accessible via mobile phones.
Result: Able to send data from sensor to Node-RED Dashboard and control relay using switch in dashboard
Browse My Recent
Tools: Keras, ResNet, Python
Description:
This project aims to classify Alzheimer’s disease stages using MRI images. A ResNet model, pre-trained on ImageNet and fine-tuned on the Alzheimer’s MRI dataset, is used to categorize images into relevant classes. The model is trained over 4 epochs to detect patterns associated with different stages of Alzheimer's.
Result:
The model achieved a validation accuracy of 97.31% by the fourth epoch, with a loss of 0.0753. This high accuracy indicates strong predictive capability, enabling more reliable classification of Alzheimer's stages based on MRI scans.
Tools: TF-IDF, Cosine Similarity, Python
Description:
This project develops a content-based recommendation system for Play Store apps, allowing users to find relevant apps based on attributes like ratings, reviews, and genre. It leverages cosine similarity for personalized recommendations without relying on user data.
Result:
The model achieved an F1-score of 82.35% with a precision of 70%, effectively recommending apps that match user preferences based on app content alone, thus improving user experience on the Play Store.
Tools: Yahoo Finance API, Prophet, Arima
Description:
This project predicts GOTO stock prices using time series models to help investors make informed decisions. Data was sourced from Yahoo Finance, focusing on price trends from 2021 to 2024.
Result:
ARIMA outperformed Prophet with lower error rates (MAE 6.11, RMSE 20.41), making it the preferred model. This approach demonstrates the effectiveness of ARIMA for short-term stock trend prediction.
Tools: TensorFlow, Google Colab
Description:
Dicoding Final Assignment for "Belajar Machine Learning untuk Pemula" course.
Result: Validation Accuracy over 98% and able to classify
Tools: TensorFlow, Scikit-learn, numpy, pandas, matplotlib
Description: Utilizing natural language processing (NLP) techniques, we categorize news articles based
on their textual content to determine whether they are authentic or fabricated.
Result: Achieve Validation Accuracy over 89%, tested using test dataset and confusion matrix
Tools: Pandas, Seaborn, Matplotlib, TensorFlow
Description:
This capstone project for the Advanced Data Analytics course involves a comprehensive
analysis of a dataset and the development of predictive models aimed at providing valuable insights to the Human Resources (HR) department of a large consulting firm.
Result: Able to see correlation and interaction between variable, gain 83% model accuracy
Tools: Pentaho, Metabase
Description: Conducting dashboard analysis of news performance in a news company, utilizing Pentaho for data processing and Metabase for visualization.
Tools: Excel, Tableau
Description: The Analytics Dashboard provides an at-a-glance view of the company's performance. This dashboard uses two metrics: net sales and new memberships. Managers use this dashboard to analyze performance and make informed decisions regarding promotions, improvements, human resources, and other actions. Please note that the data used in this dashboard is dummy data generated by AI.
Result: Give insight about relation between transaction and net sales, also performance in a month
Browse My Recent
Role: Team Leader
Tools: SPSS, Stata, Excel, Power BI
Description:
The final dashboard is themed "Dashboard Analytics for the Correlation Between Tuberculosis and Diabetes Mellitus." This analysis can be used by hospitals to take optimal actions when addressing these illnesses.
Result: Goes to final and ranked 24th out of 569 teams
Get in Touch
