Armand Blin

Master in AI — ML Research Student

Seeking final research internship (April 2026). Focus: Visual Understanding, CV, DL.

GitHubLinkedInCV

Education

Academic background and relevant coursework

MVA (Mathematics, Vision, Learning) - ENS Paris-Saclay

Sept 2025 - Apr 2026

2nd-year Master's in Mathematics, Machine Learning and Deep Learning

Paris, France

Coursework: Computer Vision, Deep Learning, Neural Architectures, Convex Optimization, ML for Time Series

EPITA - SCIA Track (AI & Big Data)

Sept 2021 - Apr 2026

Master's in Machine Learning and Data Science

Paris, France

GPA: 4.0 • Semester in SRE Team (La Forge): DevOps, K8s, Networking, Terraform • Coursework: Convex Optimization, Adv. Math for ML, Prob & Stats, ML Fundamentals, NLP

Imperial College - Mathematics for Machine Learning (Coursera)

Jul 2025 - Sept 2025

Online Specialization

Experience

Internships, research, and roles

Machina Capital

Sept 2024 - Jan 2025

Software Engineer Intern

Engineered Python libraries and migrated tools to Airflow; led big-data research to optimize trading automation infrastructure.

  • Migrated batch workflows to Airflow with clear SLAs
  • Optimized data ingestion and monitoring stack
InfraDataAirflowAWSPython

ThetysLab

Mar 2024 - Apr 2025

Lead Software Engineer (Freelance)

Designed scalable backends and cloud architectures for apps with 700K+ MAU across AWS/GCP; led full-stack and CI/CD initiatives.

  • Built multi-region deployments and observability
  • Set up CI/CD pipelines and zero-downtime releases
InfraCloudTypeScriptCI/CDGCPAWS

Peewix

2021 — 2023

Founder / Solo Venture

Built a platform for organizing and sharing events; end-to-end product development and deployment.

  • Conducted market validation and MVP shipping
  • Managed infra, backend, and mobile app integration
ProductFull-stack

Projects

Selected work

Multimodal AI Assistant — Lightweight Jarvis

Local modular AI assistant combining speech, vision, and automation tools; current work on hand-tracking for gesture interaction.

  • Speech and vision pipelines orchestrated locally for privacy
  • Experimental hand-tracking for gesture control
ML/DLMultimodalVisionPythonPyTorchOpenCVWhisper

Hybrid Recommender System for Videos (KuaiRec)

Content-based recommender for short videos on KuaiRec; feature engineering, weighting, and evaluation pipeline.

  • Built end-to-end pipeline: preprocessing → features → recommendations
  • Improved MAP and precision with tuned engagement weights
  • Best model: NDCG @500 = 0.6 - MAP @500 = 0.4
MLRecommenderPythonPandasNumPyscikit-learnJupyter

NLP — Business Reviews

Classification and generation of customer reviews; compared RNN, FFN, Transformers, Logistic Regression, Naive Bayes.

  • Trained and evaluated multiple model families
  • Reported metrics and trade-offs across baselines
NLPMLPythonPyTorchscikit-learn

Ships Classification

Kaggle competition: Classification of ships.

  • Trained and evaluated a CNN model
  • Accuracy: 82.8%
  • Ranked 17th out of 117 teams.
Computer VisionMLPythonPyTorch

Hackathons

Community & comps

Finalist — Google Healthcare Hackathon 2025

Built an AI medical assistant using MedGemma, Google Chirp, and avatars for pre-diagnosis and urgency detection.

People's Choice — Mistral AI MCP Hackathon 2025

DevOps co-pilot reducing infrastructure complexity on Mistral's LeChat.