Consumption monitoring platform for blockchain development
Project providing a ready-to-use platform for monitoring and evaluating consumption on blockchain networks.
Main goal
Platform specifically focusing on the Ethereum mainnet and a private blockchain testbench. Provide accurate and actionable data on various on-chain & off-chain consumptions, enabling users to better understand and optimize their blockchain operations.
Key Components
Ethereum Mainnet Consumption Dashboard
Utilizes client distribution and scientific reports to estimate consumption.
Integrates API data from reputable services into interactive graphs.
Estimates idle network costs, transaction costs, and daily total costs.
Private Blockchain TestBench
Deploys a private blockchain using servers as physical machines connected to wattmeters.
Implements Grafana and Prometheus for detailed monitoring of both blockchain clients and Raspberry Pi devices.
Provides an interface to configure network parameters, create and manage smart contracts, and track on-chain and off-chain metrics.
Offers easy and reliable deployment with extensive error management and configuration options.
Benchmarking LLM for Ethereum Smart Contract Development
2024
Open-source project automates the generation, compilation, analysis, and testing of smart contracts using Large Language Models with Ollama, producing statistics to evaluate model efficiency based on different prompts.
This project created at Orange-Innovation, provides an automated pipeline to generate smart contracts using Large Language Models with Ollama. The pipeline compiles these smart contracts with the Solidity compiler (solc), analyzes them using Slither, and performs unit testing based on the provided test instructions in the prompts. Statistics are then produced to determine the efficiency of each model and various prompts are used to interpret the results.
AI-enhanced development environment for solidity projects
This application consists of a web IDE enabling the integral development of solidity projects, compilation takes place on a dockeurised back-end and LLM assistance is natively integrated to enable the user to understand, develop or syntactically check the source code.
This projet run with docker-compose with this stack :
Synchronize sensors and videos in real time with different clocks (Proof of concept)
As part of a study project. It was within my reach to develop a synchronization tool connecting a
BLE Bluetooth sensor (accelerometer)
and a video acquisition: IP camera. From these sources, implementing data collection interfaces, I was able to synchronize sensor and video data
frame by frame in real time (not in post-prod). The use case was the monitoring of elevator door openings, the machine learning of boxing blows and the movement of a horse's tail during foaling.
Java CLI program capable of saving instant data in paquets in a cloud solution
Post-production preview of a sync stream
AI for Connect Four game
2023
This project aims to implement and attempt to solve a Secret variant of Connect 4.
Bot resolving variant of a cluedo, based on literal intelligence
The goal of this project is to create an AI using a logical formalization to play a variation (without board) of Cluedo. By having knowledge of known cards, unknown cards, expected cards, it is possible to have a perfect formalist beating any strategy
Formal problem solving AI :
By transforming any problem into a system based on literal clauses (see propositional logic). It is possible to apply reasoning, find a solution, or even improve a situation. This can be deduced from logical rules and formal relationships between the elements of the problem.
Screenshots :
GUI init : Sigkill to the PID of the prog to get bot data in the standard output : Result : (1000 games -> simpleVersion vs randomBot)
This topic deals with the development of an alert Website for twitter. In particular, it is requested that a user can register one (or more) words/tokens, which he wants to monitor. In case a certain threshold is being exceeded, an alert for the corresponding user should be generated. In addition, a visualization of should be presented on the Website.
Made with
React, ExpressJS & mongoDB, EmailJS, TailwindCSS
Screenshots :
Improvement and development of a PHP web application
2022
PHP application serving as a web gateway to maintenance data from a heavy proprietary software database.
Improvement and development of a PHP web application for maintenance management (Intership 2 months)
Development of a php application serving as a web gateway to maintenance data from a heavy proprietary software database only available from fixed computers.
Work carried out :
I'm a french junior Dev with a Master's degree in Computer Science and AI. I would like to specialize myself in ambitious and new projects.
Migration to PHP 8.1.4
Application optimisation : SQL requests, algorithms, switch to a MVC pattern
Management tools for technician clocking in and out
Module for the daily organisation meeting
The technicians' schedule and timetable
Reorganisation of orders, exceptional maintenance and investments
Stock & Contracts statistics
graphic redesign
Screens :
MultiAgents factory simulation
2023
Multiagents system for decentralized planification in a robotic production workshop..
Multiagents system for decentralized planification in a robotic production workshop. This makes it possible to represent skills relating to agents, a constant arrival of products in a factory and communication between agents to exchange information and goods. By creating an optimized system, we ensure that each robot performs the tasks it is best at. All using the agent-oriented library: JADE, and with a Java object architecture.
EVM (stands for Ethereum Virtual Machine) is a decentralised virtual environment that executes code consistently and securely on all nodes on the blockchain architectures mainly used to date.
Proof of Stake consensus algorith(â 30")
Blockchains : a peer to peer networks
About consensus
CAP theorem
Proof of Stake
Ethereum PoS adaptation
Epochs
Liveness
Randomness with RANDAO
A cryptographic signature scheme : BLS
Ethereum building blocks
Repartition in committees
Aggregators
Repartition in committees
Functionning
Global participation of the entire network
Sync committee
Storage : How ethereum deals with data on the blockchain(â 50")
From registry to distributed state machine
Instructions in the EVM
Gas usage
Stack-based architecture
Machine space of EVM
Stack
Memory
Account storage
Solidity and EVM code
Kind of accounts
EOA : Externally Owned Account
Contract account
Account storage
World state of the EVM
working on 256 bits (32 bytes)
Optimisation and formats
Simple : int, uint...
String, Storage order
Structures, Arrays, Mapping, Constant & immutable
Workshop with remix.ethereum (â 20")
Solidity : Improving the efficiency of gas use(â 45")
the demonstration took place in person but here are some of the commands used on a virtual machine with applications containing breaches like apache/nginx/ftp serveur and a opened socket.
Decentralized PoS (proof of stake) networks remunerate positions within validators. My automate bot makes it more convenient to access your positions and income directly from your mobile without having to connect to various pages from wallets.
Steps
Fetch various data from different API
Parse json and compute metrics
Display the data
Callable occasionally or activated every 24 hours with android widgets
This topic deal with a Java GUI program using a lot of conceptions patterns.
This Java graphical interface allows editing of a JPanel using software architecture concepts such as MVC, design patterns, and advanced data structures. The project was completed as part of learning advanced design patterns and data structures.
A nodejs script to steal ethers deposit from a hacker who tries to perform any transaction on an account that does not belong to him.
Useful to help someone who has disclosed their private key. The malicious person must deposit ether to interact with smart contracts or make transfers. Thus from his deposit the ethers are transferred and he cannot carry out his theft of the account.
Inputs thief firefox is an extension working on a firefox web browser, she takes care of retrieving all inputs with type='text', site, time and date. Once harvested, everything is sent to an online database without the user's knowledge.
Skills
AI and Machine learning
LLM, SVM Classification, regression, clustering, Multi-agent Systems, NLP, Scikit-learn, Integration of Generative AI APIs.