Shima Foolad Email & Phone Number

Ankara, Turkey

- Developed a voice AI assistant system using LangChain to handle patient inquiries, schedule appointments, and provide information in Persian. I integrated OpenAI's GPT-4o model for natural language understanding and generation. I Utilized aipa Text-to-Speech API and Whisper ASR for seamless voice interactions.- Fine-tuned a DistilBERT model from Hugging.

London, England, United Kingdom

- Developed a model to detect the age and gender of influencers’ followers based on their profile image and find the interests of followers of an influencer based on their interaction with his/her posts.

- Developed a tooth segmentation model using Ultralytics YOLOv8 to accurately isolate teeth in X-ray and RGB images.- Implemented a dental cavity detection model using YOLOv8 and image segmentation techniques to identify dental cavities. I also implemented a teeth numbering model using heuristic algorithm. I Utilized Real-ESRGAN to enhance the resolution.

• Studied deep learning architectures in detail including CNNs (AlexNet, VggNet, ResNet, GoogleNet, RCNN, Fast RCNN, Faster RCNN, Mask RCNN, YOLO,SSD), RNNs (RNTN, LSTM, GRU), GAN, Autoencoder (VAE), CapsuleNet, RBM.
• Investigated NLP such as Word2Vec and Glove embedding, all attention mechanisms, memory networks, transformer, and transfer learning models in NLP (XLNet, BERT, OpenAI GPT, GPT2, ELMo).
• Achieved In-depth knowledge of question… Show more
• Achieved In-depth knowledge of question answering models (DCN+, biDAF, MQAN, AoA, DrQA, QRN, ReasoNet, Rnet, ElimiNet, CFC,...) Show less

Khorasan-e Razavi Province, Iran

Khorasan-e Razavi Province, Iran

Doctor Of Philosophy - Phd, Artificial Intelligence

Thesis title: Improving Question Answering System in Reading Comprehension using Attention Mechanism

Master Of Science - Ms, Artificial Inteligence, 17.55 / 20 (1Rd Rank)

Activities and Societies: Thesis Title: Integrated models of bottom-up and top-down visual attention