Bassam Tarshishi

Developed an LLM-powered drawing-summary pipeline that transforms structural PDFs into structured, source-linked engineering facts. Built a RAG chatbot with hybrid search and local LLMs, and delivered FastAPI services and React/Next.js interfaces for AI workflows.

Built an end-to-end adversarial ML system for 43-class GTSRB traffic-sign recognition, including attack generation, ResNet-18 expert training, hybrid routing, and UI demo integration. Evaluated FGSM, PGD, CW, and adversarial patch attacks, achieving 93.15% mean accuracy with the strongest PGD expert.

Built a Python/FastAPI support automation system for Telegram and WhatsApp, analyzing customer messages and automatically creating Jira tickets using OpenAI-powered NLP. Integrated Jira REST APIs, PostgreSQL/SQLAlchemy, webhooks, and deployment workflows.

Graded homework, quizzes, and PSpice simulations; supported students with circuit analysis and assignments while coordinating with course instructors.

Developed an AI-assisted healthcare scheduling platform using React, Tailwind CSS, Node.js, Express, and MongoDB, with predictive scheduling, real-time updates, reminders, multilingual support, symptom checking, and auto-rebooking.

Developed LSTM models for SPY forecasting, including a general model and automated self-tuning variant; integrated FinBERT sentiment analysis and delivered predictions through a Streamlit application.

Applied Python, NumPy, Pandas, and Matplotlib to supervised ML, neural-network, and NLP exercises using TensorFlow/PyTorch. Developed a conversational AI chatbot with sentiment adaptation while participating in weekly technical mentorship sessions.

Grade: 3.86

Pursued a Bachelor of Engineering in Computer and Communications Engineering with a cumulative GPA of 3.86 and a minor in Economics with a GPA of 4.3.

Grade: A+

Completed the Lebanese Baccalaureate in General Science with a letter grade of A+.