11/10/2026 updated
AC
100 % available
AI & Embedded Engineer | Python, PyTorch, LLM, IoT/ESP32, C/C++, Flutter, React
Reims, France
Only remote
Engineering degree (MSc), Computer Science and Electronics for Embedded Systems - Polytech GrenobleAbout me
Freelance AI and embedded engineer from Polytech Grenoble, based in France and working remotely. I take products from idea to production: ML models, LLM features, IoT devices, Flutter apps and dashboards. Recent results: an AC control product with 60% lower consumption and an LLM coach used daily.
FlutterArduinoC++ (Programming Language)Embedded CEmbedded SystemsEmbedded SoftwareField-Programmable Gate Array (FPGA)Python (Programming Language)Machine LearningMessage Queuing Telemetry Transport (MQTT)Raspberry PiTensorFlowTypeScriptPyTorchReact.js
AI & Machine Learning
Expertise in Python, PyTorch, TensorFlow, Transformers, time series forecasting, model profiling and compression. Development of AI and physics-based models for energy optimization, as well as embedded AI inference on microcontrollers and FPGAs.
Embedded Systems & IoT
Proficiency in C/C++, ESP32 (MicroPython), MQTT over TLS, Raspberry Pi, Arduino, and FPGA. Experience taking IoT products from device to production, including remote AC control and energy telemetry platforms.
LLM Integration & Web Development
Integration of Large Language Models through APIs for chatbots, image analysis, and answer precision testing. Development of web dashboards using TypeScript, React, Vite, REST APIs, and Cloudflare Workers (D1, R2).
Mobile Development
Development of Flutter and Dart mobile applications, including Android widgets, Health Connect integration, and Kotlin bridges. Co-development of a fitness app with LLM-based coaching and calorie logging features.
Software & Backend
Experience with TypeScript, React, Vite, REST APIs, Cloudflare Workers (D1, R2), and Firebase Cloud Functions for building scalable serverless backends and dashboards.
FPGA & Hardware Acceleration
Research and implementation of integer LeNet inference on a Leon3 processor with the Sparrow SIMD accelerator on FPGA, achieving 3.8x faster performance. Co-authorship of a publication in Electronics (2024).
C Inference & Model Optimization
Writing C inference libraries from scratch for neural network models, shrinking models from 600K to 5K parameters while maintaining 95.3% accuracy and enabling on-device inference within 256 KB of RAM.
DevOps & Testing Tools
Familiarity with Git, GitHub, Docker, end-to-end testing, Nsight, and Valgrind for development, profiling, and quality assurance workflows.
Experience
Professional experience0 - 5 years
Freelancer since2025
Languages
GermanBasic knowledgeEnglishFluentFrenchNative speaker
Project history
Built an energy intelligence platform for supermarkets in Kenya with minute-level telemetry, 100+ dashboards, and a serverless backend on Cloudflare Workers D1 and R2. Developed a remote AC control product achieving 60% lower energy consumption. Created ESP32-S3 MicroPython device for infrared AC control with MQTT over TLS. Developed AI and physics-based models for commercial refrigeration and air conditioning optimization. Built web dashboards in React, TypeScript, and Vite with time series, heatmaps, and correlation plots.
Developed deep learning models to forecast energy consumption of client sites. Built a Flutter mobile app to control and monitor equipment. Created the first embedded version on Raspberry Pi (Python) to reduce solution cost. Designed the solution from product design to production.
Ported a Keyword Transformer (12 classes, 65,000 audio clips) to an Arduino Nano 33 BLE within 256 KB of RAM and 1 MB of flash. Wrote a C inference library from scratch validated layer by layer against Python with 100% matching predictions. Shrank the model from 600K to 5K parameters achieving 95.3% accuracy and 3.6s inference on the board.