About memy stats
Information About me
I hold a diploma of engineering from Telecom ParisTech in the field of CyberSecurity. I also hold a PhD from Sorbonne Doctoral School. During the 3 years of my PhD studies, I explored the state of the art cryptography and network security solutions. I also designed new schemes for recent applications especially for IoT devices.
Aside from theoretical knowledge, I have extensive programming expertise in a variety programming languages. I develop security solutions for network intrusions detection and secure computation over the cloud. I like to support open-source community by sharing my source code to the public.
Away from my desk, I enjoy playing chess and doing sport. My favorite sports are swimming and hiking.
Years of experience
Applied Crypto for Cloud Security
Machine Learning and AI
Virtualization (Docker, Vmware)
Unix System Administration
Debugging and Dynamic Code Analysis (Intel PIN, GDB)
Reverse Engineering and Static Analysis (IDA Pro, Radare2)
Research Engineer - THALES SIX GTS (Paris, France)
Designing and developing cutting-edge network security solutions.
Final Year Internship - Stevens Institute of Technology (Hoboken, USA)
Reducing the attack surface of user programs by removing unwanted features from programs using dynamic and static binary analysis.
Summer Internship - EURECOM (Sophia Antipolis, France)
Designing a privacy-preserving neural networks using multi-party computation.
Ph.D. - University of Sorbonne
Philosophy Degree in IoT Security (Bac+7).
Diploma in Engineering - Telecom ParisTech (EURECOM)
Diploma in cybersecurity engineering. Equivalent to a masters degree (Bac+5)
Diploma in Engineering - Lebanese University
Diploma in telecomunication engineering. Equivalent to a masters degree (Bac+5)
My ProjectsMy Work
Here is a selection of my work in several programming languages. The source code of all these projects is accessible on Github
Secure and Fault-Tolerant Aggregation
This is an implementation of the protocol presented here . The protocol aims to preserve the privacy of federated learning clients by encrypting their model updates. The encryption is additively homomorphic such that the federated learning average can be computed on the encrypted inputs.
Distributed Anomaly Detection in IoT networks
A framework for training machine learning models for anomaly detection using realtime IoT network traffic. The frameworks enables training multiple models for different types of IoT devices. It can also collect traffic generated in several networks and train in real time.
An implementation of Joye-Libert Encryption scheme for secure aggregation (defined here) This is the first and only public available implementation of the scheme.
Secret Sharing over the Integers
An implementation of the special Secret Sharing scheme which works over integers values (defined here). The scheme allows Shamir's secret sharing scheme to be used with secrets and polynomials that are not in a field.
Simulation of IoT Remote Attestation using OMNet++
This is a simulation of the protocol proposed here. FADIA is a collaborative remote attestation protocol designed to verify the software integrity of millions of devices on the network in a scalable way.
Privacy Preserving Neural Networks
Designing neural networks using secure multi-party computation. The tool enables two parties two evaluate a private machine learning model on private inputs. The details of the scheme are presented here.
Plugin for Radare2
Radare2 is an open-source reverse engineering tool. This project implements a plugin for Radare2 which serves as a clients for FIRST server. The Function Identification and Recover Signature Tool (FIRST) developed by Talos, is a framework to help reverse engineers. It makes finding similar functions easier by searching function metadata.
Benchmarks of Binary Similarity Tools
This project aims to evaluate existing function similarity techniques. It contains a database of programs, compiled for different architectures, using different compilers and several compiler flags. Using the database we benchmark the state-of-the art diffing tools.
Automated analysis of PCAP files
Conan is a network traffic analyzer that investigates pcap file, it reads the packets, reassembles all the TCP connections in the network trace, and for each connection it looks for any ambiguities.
Mohamad Mansouri , Melek Önen, Wafa Ben Jaballah, and Mauro Conti. Sok: Secure aggregation based on cryptographic scheme for federated learning (2023). To appear in PETS'23
Mohamad Mansouri , Jun Xu, and Georgios Portokalidis. Disabling unwanted functionalities in binary programs. (2023). Under revision
Mohamad Mansouri , Melek Önen, and Wafa Ben Jaballah. Learning from failures: Secure and fault-tolerant secure aggregation for federated learning (2023). To appear in ACSAC'22
Andrea Marcelli, Mariano Graziano, Xabier Ugarte-Pedrero, Yannick Fratantonio, Mohamad Mansouri , and Davide Balzarotti. How machine learning is solving the binary function similarity problem (2022). In Usenix (Ed.), Usenix 2022, 31st usenix security symposium, 10-12 august 2022, boston, ma, usa, Boston. Retrieved
Mohamad Mansouri , Wafa Ben Jaballah, Melek Önen, Md Masoom Rabbani, and Mauro Conti. FADIA: fairness-driven collaborative remote attestation (2021). WiSec '21: Proceedings of the 14th ACM Conference on Security and Privacy in Wireless and Mobile Networks
Contact me here
Do you have an offer for me?! I'm very interested in knowing more about your offer. Please don't hesistate to contact me.
mohamad_mansouri (at) outlook.com