Minas Sifakis

  • Machine Learning and
  • Acoustics

About Me

I am a physicist turned machine learning scientist. I have experience in time series forecasting, statistical modeling, unsupervised clustering, anomaly detection and natural language processing (NLP). I hold a degree in physics, an MSc in Computational Statistics and Machine Learning from UCL and an MSc in Sound and Vibration Studies from the University of Southampton.

I am currently a research associate at the University of Edinburgh, where I work on NLP and multilingual document matching and alignment at petabyte scale. Before that I was at the Alan Turing Institute, where I used a combination of pre-trained language models, transfer learning, search engine technologies (Elasticsearch) and a vector similarity engine (FAISS) to develop a state of the art system for large scale news article monitoring, fraud detection and question answering.

I have also worked in finance (Cisapline Ltd), developing algorithms for forecasting financial time series using neural network and Gaussian processes and in ML consulting (Machinable), developing ML-based transformation strategies for large corporates. Before my switch to machine learning I had worked for ten years in engineering acoustics, two years in academia (full-time lecturer) and four years in risk management.


Natural Language Processing

I am currently working on NLP. I have experience in natural language based information retrieval, document similarity and clustering, text extraction from pdf documents, etc.

Time Series Forecasting

I have worked in time series forecasting, using

signal processing methods, Neural Networks, SVMs,

Gaussian Processes and tree based methods.

Computer Vision

I am involved in two satellite imagery based computer vision projects.

The "Slum World" project, where we try to identify slums areas in cities,

and the "Industry Relocation" project, where we try to estimate air polution from space.

Anomaly Detection, Density Estimation

I have experience working with Variational Autoencoders,

normalizaing flows and Bayesian models for anomaly detection,

probability density estimation, uncertainty quantification

and synthetic data generation.


I have 15 years experience in engineering acoustics,

noise and vibration control.


Python, Matlab, Tensroflow, Pytorch

GP-Flow, Elasticsearch/Kibana

Docker, Git, Bash, SLURM, AWS.

Other stuff

Publications, public speaking, etc.

Get In Touch

Thank you! Your Message has been sent. Something went wrong, Please try again!