Computer Vision

Time series

  • BLSTM Segmentation Model: Sequential segmentation model, using BLSTM, implemented with TensorFlow.
  • LSTM Classification Model: Classification model, using LSTM, for sequential data of various length, “last frame” prediction, implemented with TensorFlow.

NLP / Computational linguistics

Social Media Botting/Listening

  • Vitaminst, Instagram bot: I created a smart instagram bot able to target with high precision posts and create relevant comment in order to generate a feedback. Up to 250 new followers per day, deployed on 5 Raspberry Pi 3.
  • CredTracker (On going): I want to create a simple automation tool that would estimate one’s population based on tweets over time, and backtest it with official polls.

Retail

  • Wifi-based flow analysis (Company project): Designed and implemented a solution to collect, enrich and aggregate wifi location data in order to give insights to retail owners about key indicators such as, shops’ attractiveness, customers’ habits, live flows.
  • Shopping mall affluence prediction (Company project): Tested different LSTM classification models in order to predict the number of distinct customers of a shopping mall for a given day. Integrated OpenData features such as weather (+ weather forecast), public holidays, shopping mall events.

IOT

  • Building hypervisor (Company project): Designed and implemented IOT data collection worfklow + exposition worfklow in order to demonstrate key indicators for buildings owners, occupation / comfort / energy consumption etc..

Advanced Algorithms

Competitive Programming

  • ACM ICPC solutions: My different solutions for the past problems of ACM ICPC. https://icpc.baylor.edu/worldfinals/problems
  • Codingame solutions: My different solutions for the Codingame platform. https://www.codingame.com/