[info] Internship Opportunity: ML Speech Enhancement / Source Separation at Bose Hear Research

[Ignatius S. Condro Atmawan B.]

-------- Weitergeleitete Nachricht --------
Datum:  Thu, 16 Jan 2020 13:20:14 -0800 (PST)
Von:    Marko Stamenovic 

Hi All,


Bose Hear Research is currently accepting applications for 2020 interns
and co-ops in machine learning with applications to speech enhancement
and source separation. The opportunities are a full-time, paid, three to
six-month positions ideally starting in the Spring/Summer and are be
based at Bose's brand new offices in Brighton, MA right across the river
from Harvard Sq. *The candidate must be eligible to work as an intern in
the US.*


If interested, please apply directly to this req Machine Learning /
Neural Signal Processing Internship
<https://boseallaboutme.wd1.myworkdayjobs.com/en-US/Bose_Careers/job/US-MA---Boston-Landing/Machine-Learning---Neural-Signal-Processing-Internship_R17200
>. 


Bose <https://en.wikipedia.org/wiki/Bose_Corporation> is a privately
held corporation with a rich history of research and innovation in DSP,
acoustics and beyond. The majority of its shares are owned by the
Massachusetts Institute of Technology in a unique blind trust. Bose
recently got FDA approval
<https://techcrunch.com/2018/10/05/bose-hearing-aid-gets-fda-approval/>
to sell hearing aids directly to customers in a ‘De Novo’ grant and is
vigorously pursuing research & development in this area.


Also feel free to send your resume and cover letter and any questions
directly over to me at marko_stamenovic <at> bose.com <http://bose.com>.


Thanks!

Marko Stamenovic

Machine Learning Engineer



*Description:*


The goal of our team is to explore new ways that allow individuals with
hearing loss to hear better in noisy environments. Our work centers on
implementing and testing cutting edge deep learning algorithms for tasks
such as source separation and speech enhancement. The twist is, our
algorithms have to run on live audio, in real time, on real devices. As
part of the team you will be working with experts in machine learning,
deep learning, digital signal processing and psychoacoustics to
prototype and implement new algorithms. 


Bose has a strong history of combining creative thinking with cutting
edge technology in the audio domain. We’re looking for candidates
passionate about ML and audio to help shape Bose’s next chapter! 


*Responsibilities: *

  * As an intern on our team, a large chunk of your time will likely be
    devoted to prototyping and implementing ML and software-based
    solutions and experiences. You will investigate novel algorithms,
    analyze and evaluate them against big data sets, help curate and
    develop large internal audio datasets, and present your results. 
  * In addition, you will integrate solutions into new and existing
    systems and platforms to show new proof of concepts. 
  * A big part of your job may also include researching and implementing
    model compression techniques to deploy algorithms on hardware and
    edge platform


*Education: *

  * Pursuing or just finished a degree in ML, Computer Science or a
    related field. Graduate level candidates preferred, but the position
    is really open to any student or recent grad with the right interest
    and skillset. 
  * Candidate should be familiar with material covered in introductory
    courses on linear algebra, statistics, digital signal processing and
    data structures



*Skills: *

  * Able to develop in Python, Tensorflow and/or C/C++ (just R or Matlab
    won’t cut it) 
  * Strong understanding of Machine Learning and DSP concepts 
  * Familiar with standard development practices (version control,
    testing, micro-
    services) 
  * Strong communication skills and comfortable with presenting analyses
    to a wider
    group (you will have the opportunity to present your work on a
    regular basis). 
  * Familiarity with the some of the following technologies is also
    preferred: TensorFlow, TFLite, CoreML, AWS, Git, Jupyter Notebooks,
    etc. 
  * Extra points for familiarity with neural audio synthesis, source
    separation, model
    compression & pruning. 
  * Interest/experience with implementing machine learning algorithms on
    hardware or edge devices a BIG plus. 

--
Open-access journal Transactions of ISMIR, open for submissions:
https://tismir.ismir.net
---
ISMIR 2019 will take place in Delft, The Netherlands, November 4-8, 2018.
Website: http://ismir2019.ismir.net/
---
ISMIR 2020 will take place in Montréal, Canada,
ISMIR Home -- http://www.ismir.net/
---