Dr Heeba Altaf gurku

Innovative Machine Learning Engineer with a robust background in medicine, specializing in radiation oncology. Specialized in developing advanced ML/DL models and solutions, including creating a medical large language model that surpasses Med-PaLM in benchmark datasets like USMLE QA and outperformes other LLMs in human evaluation tests. Proficient in Python, predictive and generative models, NLP, prompt engineering, and data analysis. Experienced in implementing ML solutions for medical appeals using LLMs. Adept at collaborating with cross-functional teams, driving projects from conception to deployment, and integrating domain knowledge with ML to tackle complex problems and deliver impactful results. Despite a strong medical background, I have also worked on small projects in other domains and am keen to expand my expertise into finance, automation, biotech, and more.

CERTIFICATIONS Professional Data Science Certificate, IBM-Coursera Python 3 Programming Specialization, University of Michigan-Coursera Basic Statistics, University of Amsterdam-Coursera

• English: Fluent
• Hindi: Fluent

• Sparks Investment Analysis: Conducted exploratory data analysis
• Lending Club Study: Performed univariate, bivariate, and multivariate analysis
• Australian Housing Market Analysis: Applied linear regression with regularization
• Bike Sharing: Developed a linear regression model
• Telcom Data Analysis: Implemented logistic regression, decision tree, random forest, boosting, PCA, hyperparameter tuning, and SMOTE
• Melanoma Detection: Built CNN models using TensorFlow and Keras
• Gesture Recognition: Utilized CNN, RNN, LSTM, GRU, and transfer learning
• Healthcare Data Entity Identification: Employed NLP techniques with Spacy and CRF
• Automatic Ticket Classification: Implemented NLP and topic modeling
• Tic Tac Toe Game: Developed a reinforcement learning model
• Cab Driver Assistance System: Created a deep Q-learning system
• MRI T1/T2 Image Style Transfer: Utilized Cycle GAN
• Synthetic CT Image Generation: Applied Cycle GAN and Attention Guided GAN
• Clinical Subspecialty Prediction: Fine-tuned BERT, RoBERTa, XLNet, and BioBERT on medical transcription notes.