Engineering student · ISEN Nantes · Class of 2027

Ewan Menaheze

Final-year student at a French graduate school of engineering (computer science and electronics), specialising in artificial intelligence. I build computer vision and deep learning models, from data preparation all the way to a working demo.

Nantes, France

Portrait of Ewan Menaheze

About

What drives me

I am passionate about artificial intelligence, especially deep learning and image processing.

Alongside my studies, I take on projects to put these fields into practice: tracking and re-identifying people in video, embedded vision on an electronics board, large-scale maritime data analysis. My research internship at the University of Coimbra, on the analysis of capsule endoscopy videos, confirmed this direction. Every project shown here is documented, from the approach to the results.

School
ISEN Nantes
Degree
Engineering degree 2021 – 2027
Major
Artificial intelligence
English
C1 level

AI and vision

  • Deep learning
  • CNN
  • Transfer learning
  • Object detection (YOLO)
  • Image processing
  • OpenCV
  • TensorFlow · Keras
  • PyTorch

Data

  • Python
  • R
  • scikit-learn
  • Clustering
  • Data visualisation
  • SQL

Software

  • C++
  • Java · JavaFX
  • HTML · CSS · JavaScript
  • PHP
  • REST APIs
  • Git
  • Embedded: Arduino · BeagleBone · FPGA

Day to day

  • Teamwork
  • English C1
  • French (native)

Contest · 2026 · 8th edition

Euro-Information coding contest

“For a responsible AI”: a 52-hour optimisation challenge, solo, with AI tools allowed. The task was to choose which AI models a computing centre should train under an energy cap to maximise their total value, across seven input files of up to 50,000 models.

  • 14thout of 177 entrants
  • 52 hof competition
  • 7input files
The problem, the seven files and the method

From 9 October (8 pm) to 11 October 2026 (11:59 pm). A single problem split into 7 input files: one answer file is submitted per input, and each is worth up to 1,000,000 points, pro rata to the best score in the contest.

Each model needs an amount of data of 4 types (numerical, text, images, code) within a given range. Each dataset can be used only once. A copyrighted dataset halves the model's value. A model that uses a single data type can act as a generator to cover another model's need, and that model then also loses half its value.

  1. “Example” (5 models, 5 datasets): a tiny test case, solved to optimality (12,946).
  2. “So it begins.” (3,000 models, 10,000 datasets): copyright-free data is scarce, and models that need only a sliver of data “waste” a whole dataset. Best score: 272,318.
  3. “Everything is free!” (5,000 models, 20,000 datasets): everything is copyright-free, so it is a pure knapsack problem. Proven optimum (523,758).
  4. “Those models are demanding!” (8,000 models, 40,000 datasets): very narrow data ranges (10 units at most) and no generators allowed. Each need has to be filled almost to the exact dataset. Best score: 1,031,941, at most 3 points from the optimum.
  5. “Looks like there's a discount on our electricity bill.” (11,000 models, 100,000 datasets): energy is plentiful; data is the limiting factor. Best score: 3,786,059.
  6. “Where are the files?” (15,000 models, only 5,000 datasets): most of the value comes from models fed solely by generators, and every dataset counts. Best score: 669,670.
  7. “We're in the endgame now.” (50,000 models, 250,000 datasets): the largest file, with every mechanism at once. Best score: 5,822,547.

Method. Integer linear programming (HiGHS, Gurobi) to choose the models, then a C++ allocator (local search and exact subset sums) to assign the datasets. Several Claude sessions worked in parallel, each on one file.

Loophole left alone. The official grader accepted negative indices, which counted some models twice. Some competitors took advantage of it; I did not use it, and the organiser later fixed it.

Award · 2025

Euro-Information coding contest

Four algorithmic challenges to solve against the clock, competing with 307 participants. The result earned a €150 prize.

  • 2ndout of 307 entrants
  • 1stin C++ (22 entrants)
  • 4challenges solved
The four challenges
  1. String analysis and frequency counting with hash tables, for linear-time counting.
  2. Shortest path in a weighted graph, solved with Dijkstra's algorithm and a priority queue.
  3. Dynamic programming with state compression, to meet the time and memory constraints.
  4. Greedy interval scheduling: sorting, then conflict resolution to maximise throughput.

Work

My projects

Five pieces of work, from a research internship in medical imaging to embedded electronics. Each page details the approach, the tools and the results.

Research internship · 4 months

Automated reading of capsule endoscopy videos

Research internship at the University of Coimbra: a deep learning pipeline that finds lesions in capsule endoscopy videos, names them and locates them in the digestive tract, evaluated on patients the models have never seen.

  • PyTorch
  • EfficientNet
  • Grad-CAM
  • OpenCV
  • scikit-learn

M1 project · Computer vision

Assessing a person's cooperation for face recognition

M1 project carried out by a team of two: a system that assesses, from video, whether a person cooperates with a face recognition device. It tracks and re-identifies each person, analyses their gait, and measures face occlusion and gaze direction.

  • YOLOv8
  • BoT-SORT
  • InsightFace
  • MediaPipe
  • OpenCV

Big Data · AI · Web

Maritime traffic analysis from AIS data

Millions of vessel positions from the Gulf of Mexico: cleaning and statistics, trajectory clustering, prediction of vessel type and position, all served through a web application.

  • Python
  • R
  • scikit-learn
  • PHP · SQL
  • JavaScript

Embedded electronics

Electronic tic-tac-toe

A tic-tac-toe game for children, built by a team of four in one month: nine light sensors recognise the tokens, LEDs and an LCD screen display the game.

  • Arduino
  • C++
  • 3D printing
  • Laser cutting

Contact

Get in touch

A question about a project, an opportunity, or simply want to talk? Write to me with the form or directly by email.