Building a CPU-optimized LLM, Retrieval-Augmented Generation (RAG) systems and a voice-to-text summarizer for internal use. Working under CPU-only hardware constraints, which drives the focus on quantization and efficient local inference.

• LLMs
• RAG
• CPU inference
• Speech-to-text

Assisted in building and evaluating machine learning models on real-world datasets. Performed data cleaning, feature engineering and exploratory data analysis. Implemented classification and regression models with Scikit-learn and evaluated them with accuracy, F1-score and ROC-AUC. Improved performance through hyperparameter tuning, documented experiments, and presented findings to senior developers.

• Python
• Scikit-learn
• Pandas
• NumPy
• Matplotlib
• Seaborn
• PyTorch
• TensorFlow

Built backend systems and AI-driven applications. Designed and implemented scalable web services with FastAPI and Django REST Framework, integrated databases, and managed authentication and API endpoints. Developed chatbots and intelligent applications using LLMs through Hugging Face and OpenAI APIs, applying prompt engineering and safety filters for reliable responses. Worked on predictive modeling (for example house price prediction) deployed with Streamlit. Debugged and optimized ML pipelines and helped keep applications performant and secure.

• Python
• FastAPI
• Django REST Framework
• Hugging Face Transformers
• OpenAI GPT models
• Prompt engineering
• Pandas
• NumPy
• Regression modeling
• Streamlit
• SQL/ORM
• Docker (basic deployment)
• Git & GitHub