Posted 3 months ago
About this position
Jukedeck is an AI tech startup using artificial intelligence to personalise music. We’re a team of composers, producers, researchers and developers with backgrounds that span big tech companies, academia and the music industry, and we’re based in London.
We’re looking for a Junior Data Science Intern to help speed up the R&D team workflow. This will include dataset management, model training and selection, qualitative and quantitative evaluation of machine-generated music, documenting experiment results and managing training resources.
You’ll be responsible for helping other machine learning researcher with their experiments and make sure the flow of experiments runs smoothly. You will be able to learn many useful hands-on and practical skills that will aid you in becoming a successful machine learning researcher. These skills would be transferable across various machine learning application domains and will teach you to thoroughly and efficiently carry out machine learning experiments, document your observations, analyse results and plan future experiments.
Your responsibilities will include:
- Examining and cleaning musical data in MIDI format.
- Vectorising MIDIs in different numerical formats to train machine learning algorithms.
- Training and selecting machine learning models (locally and on remote servers).
- Generating music with the trained models.
- Carrying out a quantitative and qualitative analysis of the generated music.
- Summarising and discussing your analysis with other machine learning researchers.
- Making recommendations regarding next steps.
- Meticulously documenting various experiment details, organising data, experiment settings, trained models, generations and statistics
- Perform simple git related tasks such as pull request management for datasets and new models
- Managing various server administration tasks, both for Amazon AWS servers and our internal server
- Writing simple reusable Python/shell scripts for batch-processing data and data/results analysis.
At Jukedeck, we embrace diversity and foster a fair and inclusive culture. We encourage everyone to apply, regardless of gender identity, sex, sexual orientation, race, color, religion, national origin, disability and age. We would like to make the future of music and AI diverse and inclusive – join us in this mission!
Our ideal candidate will:
- Be an undergraduate student who has completed courses in machine learning or a data science masters programme student looking to apply what she/he is learning in the course
- Be familiar with most common ML concepts, neural networks and deep learning, (in particular recurrent neural networks), model selection, overfitting, regularisation, training/validation, etc.
- Be familiar with machine learning based approaches for composing music.
- Have understanding of the Linux command-line, in particular commands related to the SSH protocol
- Have a good knowledge of basic Python programming and Tensorflow
- Have some familiarity with AWS server usage and be familiar with version controlling with Git
- Be very well-organised, have good time management skills, is thorough at documenting her/his work.
- Be keen to work in a startup environment.
- Be able to work in our London office.
Your application should include:
- Your CV and a link to your LinkedIn profile
- Any of your other work (publicly shared code, past projects, demos, music) you’d like us to know about