Ulviyya Aliyeva
AI Engineer & Data Scientist
An enthusiastic AI Engineer and Data Scientist with a strong interest in NLP, building ethically aligned AI systems and fostering learning.
Haqqımda
Enthusiastic and curious AI with a strong interest in Natural Language Processing (NLP) and Data Science. Passionate about using artificial intelligence to improve human lives. Continuously learning and growing to one day contribute to the development of human-like, ethically aligned AI systems.
İş Təcrübəsi
Junior AI engineer
Kontak Home
Building "Bots" /RAG systems. Research Ai solutions for automation manual works.
- Built "Bots" /RAG systems.
- Researched AI solutions for automation of manual tasks.
AI Engineer Intern
Kapital Bank
Assisted in the implementation of the "Aila" RAG chatbot, specifically by adapting workflow logic and Python scripts to support multilingual functionality. Validated system guardrails by testing edge cases and refining prompts to ensure safety across different languages. Gained hands-on experience with Dataiku and LangGraph by troubleshooting and optimizing specific components of the pipeline.
- Assisted in implementing 'Aila' RAG chatbot, adapting workflow logic and Python scripts for multilingual functionality.
- Validated system guardrails by testing edge cases and refining prompts for safety.
- Gained hands-on experience with Dataiku and LangGraph, troubleshooting and optimizing pipeline components.
Beta-tester & Mentor
DeepLearning.ai
Selected as a beta tester for new courses, features, and tools developed by DeepLearning.AI, a leading platform for AI and machine learning education. Provided detailed feedback on the content quality, technical clarity, and usability of learning modules. Tested hands-on exercises and coding environments, identifying bugs or unclear instructions. Assessed model outputs and DL workflows (especially in NLP-focused content) to ensure pedagogical effectiveness. Contributed user-centered insights to improve learner experience. This role allowed me to deepen my understanding of deep learning concepts while supporting the improvement of educational tools for a global AI learning community.
- Selected as a beta tester for new DeepLearning.AI courses, features, and tools.
- Provided detailed feedback on content quality, technical clarity, and usability of learning modules.
- Identified bugs and unclear instructions in hands-on exercises and coding environments.
- Assessed model outputs and DL workflows for pedagogical effectiveness, especially in NLP.
- Contributed user-centered insights to improve learner experience and educational tools.
AI Engineer/Data Science Mentor
Div Academy - Baku, Azerbaijan
Supported students in understanding core concepts of Data Science, with a focus on: Python programming and libraries (pandas, NumPy, matplotlib), Data cleaning, preprocessing, and basic feature engineering, Introductory machine learning algorithms (e.g., linear regression, KNN, decision trees using scikit-learn), Foundational tasks in Computer Vision (e.g., image preprocessing, simple classification). Passionate about AI and Natural Language Processing (NLP), and actively work on improving my skills in these areas. While I help students build confidence in their technical abilities, I also remain open about the fact that I'm still learning and growing - especially when it comes to more advanced topics. For me, mentoring is not about having all the answers - it's about learning together, encouraging curiosity, and building a supportive space where students (and mentors!) grow through shared exploration and hands-on problem solving.
- Mentored students in core Data Science concepts, including Python, data preprocessing, machine learning algorithms, and Computer Vision.
- Focused on Python programming, pandas, NumPy, matplotlib, data cleaning, feature engineering, linear regression, KNN, decision trees, scikit-learn, image preprocessing, and simple classification.
- Fostered a supportive learning environment, encouraging curiosity and hands-on problem-solving.
Data Analyst Intern
DataVision
Analyzed datasets, generated insights. Created visualized trends. Cleaned data, ensured accuracy. Collaborated. Hands-on experience in sklearn, numpy, pandas. Utilized data tools to optimize workflows.
- Analyzed datasets and generated insights.
- Created visualized trends.
- Cleaned data and ensured accuracy.
- Gained hands-on experience with sklearn, numpy, and pandas.
- Utilized data tools to optimize workflows.
Bacarıqlar
AI/ML/Data Science
Programming Languages
Libraries/Frameworks
Databases
Tools/Software
Soft Skills
Layihələr
Team Project for educational purposes, designing and implementing a decoder-only Transformer (GPT) architecture from scratch, including Causal Multi-Head Attention, Feed-Forward MLP blocks, Layer Normalization, Residual Connections, and Dropout. Applied weight tying, developed training loop with gradient accumulation, AdamW optimizer, cosine learning rate scheduler, and gradient clipping. Built a lightweight DataLoader and implemented autoregressive text generation with top-k sampling. Leveraged insights from GPT-2/3 papers, Attention Is All You Need, and Andrej Karpathy's nanoGPT tutorials. Actively working on model evaluation and optimization.
Conglomerate Concrete Crack Detection (Automated Concrete Crack Segmentation using U-Net Architecture)
Developed an end-to-end deep learning pipeline for concrete crack detection using Keras and TensorFlow. Created a custom data loader with image and mask pairing, normalization, resizing, and augmentation. Implemented a U-Net architecture with skip connections for precise pixel-wise segmentation. Developed custom loss metrics (Binary Cross-Entropy + Dice loss) and Dice Coefficient. Trained the model on a large dataset, achieving significant improvement in segmentation accuracy. Created a visualization and overlay system. Utilized advanced training techniques such as data augmentation and parallel data pipelines.
Developed a character-level text generation model using TensorFlow and Keras. Created a custom data preprocessing pipeline, including character-based tokenization and sliding window generation. Trained a GRU-based neural network for next-character prediction. Implemented temperature-controlled sampling for diverse text generation. Evaluated model performance with validation sets and used checkpointing.
Developed a system to recognize hand-written digits shown via fingers and control a computer interface. Built and trained a custom Convolutional Neural Network (CNN) from scratch. Implemented most logic manually, integrating MediaPipe for hand detection and landmark tracking. Focused on end-to-end pipeline: gesture capture → hand detection → preprocessing → digit classification → interface control.
Developed a live webcam emotion detection application using OpenCV for face detection and Keras for deep learning-based emotion classification. Leveraged transfer learning by fine-tuning the Xception convolutional neural network on an emotion dataset. Created an end-to-end inference pipeline achieving high accuracy and responsiveness. Trained and saved the model in .keras format for seamless deployment.
Təhsil
Div Academy
Student — Data Science
Azerbaijan State University of Economics • UNEC
Master's degree — Expertise and marketing of food products
Azerbaijan State University of Economics • UNEC
Bachelor's degree — Food engineering
Sertifikatlar
Neural Networks and Deep Learning
Coursera
Machine Learning
Coursera
Advanced Learning Algorithms
Coursera
Unsupervised Learning, Recommenders, Reinforcement Learning
Coursera
Supervised Machine Learning: Regression and Classification
Coursera
1st place-ClimaTech and Sustainability Hackathon
Impuls4Women, IDDA
1st Place Winner - Brand Boom Marketing Competition
Azersun Holding
B2 English language course certificate
Girne American University (Cyprus)
Python course
BP
