Kseniia Lider

SUMMARY

Transitioning to Machine Learning with over 10 years of diverse experience in tech, including roles in software quality assurance, project/product management, and chatbot development. Holds a Master's in Control Engineering with peer-reviewed publications. Hands-on experience in Python, data analysis, predictive modelling, machine learning frameworks and LLMs. Seeking a role in Data Science, Machine Learning or AI to leverage my vast tech background, result-oriented entrepreneurial mindset and drive for intellectual challenge to deliver robust solutions to complex real-world problems.

PROJECTS

Building a RAG system for technical standards

• Used LlamaIndex library to implement the pipeline of parsing, chunking, embedding, retrieval and context-augmented query processing.
• Built a Streamlit web interface with real-time English/Ukrainian language switching.

Developing an application for digit recognition

• Trained a convolutional neural network (CNN) in PyTorch on the MNIST dataset and served the model via a FastAPI web server.
• Logged predictions to PostgreSQL and containerised the solution with Docker.

Fine-tuning LLM for grammar correction

• Fine-tuned a pre-trained model with a llama-style architecture for grammatical error correction.
• Utilising the Hugging Face libraries, applied Supervised Fine-Tuning (SFT) on the CoEdIT dataset and aligned the model using Direct Preference Optimisation (DPO).

Building models for electricity price forecasting

• Completed the Energy Academy course and built ML models to predict electricity price spreads between day-ahead and intraday markets.
• Explored data using visualisation (Plotly), performed data preprocessing (imputation, feature engineering), and applied classical ML algorithms with hyperparameter tuning (Scikit-learn).