DermSense Classifier
A medical imaging neural pipeline contrasting an SVM-HOG baseline with a PyTorch EfficientNet-B0 CNN. Integrates Grad-CAM visual heatmaps for diagnostic transparency.
from langchain import OpenAI, AgentExecutor
from tools import WebSearch, PythonREPL
class AIEngineerAgent:
def __init__(self):
self.name = "Charbel Toumieh"
self.role = "AI & LLM Engineer"
self.tools = [WebSearch(), PythonREPL()]
def solve_complex_task(self, prompt):
agent = self.init_llm_chain()
return agent.run(prompt)
# Instantiating professional intelligence
ct = AIEngineerAgent()
ct.solve_complex_task("Build production-grade AI")
I am an AI & LLM Engineer specializing in building production-grade intelligent systems and agentic workflows. Currently pursuing my MSc in Applied Artificial Intelligence at the University of Warwick, I bridge research methodologies with production deployment. My background includes **Meta-funded AI collaboration** on multi-modal neural systems and deploying **production LLM chatbots** that achieved 25% efficiency gains at Gabriel & Co. I focus on RAG pipelines, multi-agent frameworks, and scalable backend microservices.
Designing multi-agent tool loops & self-correction
Deploying robust, low-latency container APIs
Hands-on engineering across US, UK, and Europe
University of Warwick
American University of Beirut
Developed curriculum and lectured a foundational AI course at AUB
May 2024 – Sep 2025Guided activities and initiatives, fostering professional development
Feb 2022 – May 2023Actively facilitated daily coding meetings, contributing to Python codebase
Sep 2021 – Jun 2022
A medical imaging neural pipeline contrasting an SVM-HOG baseline with a PyTorch EfficientNet-B0 CNN. Integrates Grad-CAM visual heatmaps for diagnostic transparency.
Student retention pipeline following CRISP-DM. Modularized into feature selection (XGBoost RFECV), classifiers/regressors, and unsupervised clustering.
A network metrics reporter capturing LTE/5G/Wi-Fi telephony parameters. Transmits telemetry data via Flask APIs to a local database system.
Cross-platform mobile and desktop application targeting multiple OS configurations, powered by .NET MAUI Blazor and offline SQLite storage.
FastAPI backend microservices architecture. Features segregated customer/order databases, HAProxy load balancing, and Prometheus execution scraping.
Mathematical modeling simulations representing Discrete Fourier Transforms, Nyquist sampling limits, Uniform/Selective quantization noise, and coherent BPSK/QPSK modulation.
A path-finding logistics simulator modeling layouts as weighted graphs. Implements graph algorithms (DFS/BFS) for routing, quicksort, and transaction rule analytics.
Interactive dashboard for visualizing and analyzing public health data with dynamic charts and data-driven insights representing global COVID telemetry.
A feedback aggregation tool rendering statistical distributions and predicting wellbeing patterns via relational SQL.
Machine learning pipeline utilizing CDC telemetry statistics to classify and cluster diabetes indicators.
End-to-end cryptographic web application implementing classical ciphers (Affine, Vigenere, Hill) deployed on Azure.
Python module utilizing Matplotlib and Seaborn for dynamic plotting of customized datastreams.
A web system containing custom administrative tools, integrated messaging systems, and OpenAI Chatbot configurations.
A library management system implementing core functionality for book and member management via OOP.
Python test-coverage assertions integrated with a Jenkinsfile pipeline automating container deployment and quality validation.
Contact management application with CRUD operations, search functionality, and data persistence.
Low-level C++ console implementation of graph traversal algorithms (Dijkstra, MST) and weighted adjacency matrix alterations.
VHDL design modeling hardware arithmetic logic gates, control selectors, and execution testbenches.
Low-level socket programming representing a multi-client HTTP proxy server with connection routing, data parsing, and caching logic.
Multiplayer game implementation with network delay simulation and synchronization mechanisms.
Flask-based enterprise employee registry and task scheduler backed by local SQLite storage.
Desktop GUI application built with PyQt framework demonstrating modern UI/UX patterns.
RESTful user registry API served via Flask with cross-origin resource sharing (CORS) security.
I am available for AI development opportunities, consulting on LLM agents/pipelines, or technical collaborations. Let's discuss your next breakthrough project.