Research Software Engineer

Vacancy Ref: : 053228 Closing Date : 15-Oct-2020

Contact Person: Jenny O’Donnell Contact Number: 01316502723

Contact Email:

The School of Informatics, University of Edinburgh, invites applicants for a full-time Research Software Engineer. We drive impact through widely-used open-source software and open data mined from a petabyte of the web. As a research software engineer, you will be making software more usable, faster, and more reproducible publicly on GitHub with partners and other users.

The position is with the machine translation group at the University of Edinburgh, which plays a leading role in the translation community. Prior experience with machine translation is not required; this is an opportunity to gain natural language processing knowledge and experience by leveraging your existing C++ skills.

Alongside researchers, other software engineers, students, and our project partners, you will work on projects like:

  • Improving the documentation of the open-source Marian toolkit for fast translation in C++ through examples, tutorials, and operator specification, or changing the toolkit to make it more malleable
  • Participating in or organizing machine translation speed contests like
  • Reproducibly building and distributing models for a Firefox extension that does client-side translation using Marian
  • Mining the web for translations in low-resource languages

This is a full-time, fixed-term appointment available until December 2021.

Grade scale: UE06 (£28,331 – £32,816pa)

Inquiries should be directed to Dr. Kenneth Heafield:

Feedback will only be provided to interviewed candidates.

The closing date is 5 pm (GMT) on 15th October 2020.

Job Details

Job title: Research Software Engineer
School: School of Informatics
Unit: Institute for Language, Cognition, and Computation
Line manager: Principal Investigator

Job Purpose

Drive the impact of machine translation research by improving software and expanding our efforts to additional languages.

Main Responsibilities

  • Make it easier to do research that modifies the Marian toolkit by improving documentation, writing examples, making tutorials, and making it more malleable. (40%)
  • Build and experiment with translation systems using Marian, initially with assistance from others in the group. (20%)
  • Demonstrate software and data to potential users at conferences, site visits, or online meetings. Contribute to research publications with other team members to maximize the impact of the project. (20%)
  • Provide feedback on the methodology and contribute to design and implementation, to identify problems or topics for future research. (10%)
  • Maintain software, identify issues, repair defects, and contribute to regular software releases, to demonstrate good research software engineering practice. (10%)

Planning and Organising

The post holder will agree to a general plan of activity with the PI and progress will be monitored at scheduled meetings.

There will be an opportunity to shape some of the direction of the project based on your interests.

Problem Solving

We want Marian to be fast and usable. Sometimes the best way to accomplish this is to change how parts of Marian work.

Decision Making

The post holder will determine the specific techniques and approaches to be used in the course of the research, in agreement with the PI.

Key Contacts/Relationships

User-Focused Marian is an EU-funded collaboration between The University of Edinburgh, Unbabel, and Tilde. You will primarily work with Marian developers, some based at Microsoft and some at the University of Edinburgh. As this position emphasizes open research, you will also be interacting with users who have questions.

Knowledge, Skills, and Experience Needed for the Job


  • Significant software engineering experience
  • Strong C++ programming ability
  • Familiarity with neural networks
  • Ability to communicate complex information clearly in English, orally and in writing


  • Track record of open-source scientific software development and documentation
  • Experience with Docker
  • Knowledgeable with git
  • Experience in machine translation or natural language processing


There are no line management responsibilities associated with this post.

  • Job Context and any other relevant information

    This post is primarily funded by the User-Focused Marian project ( though, depending on interest, you may collaborate on mining the web for translations ( or distributing fast Marian as a Firefox extension (

    Application Procedure

    All applicants should apply online by clicking the apply link at the bottom of this page and submitting an up to date CV. The application process is quick and easy to follow, and you will receive an email confirmation of safe receipt of your application. The online system allows you to submit a CV and other attachments.

    The closing date is 5 pm on 15th October 2020.

    Right to Work

    In accordance with the Immigration, Asylum and Nationality Act 2006 and Immigration Act 2016 the University of Edinburgh, as an employer, has a legal responsibility to prevent illegal working and therefore must check that all employees are entitled to work in the United Kingdom (UK).
    To do so, the University requires to see original documents evidencing the right to work in the UK before the commencement of employment, and this is normally carried out at the interview. Details will be provided in any letter of invitation to interview.

    For further information on the right to work please visit our right to work webpage

    If You Are From Outside The EEA And Not Currently Eligible To Work In The UK, There Are Visa Routes That May Be Available To You, For Example

    • Global Talent: If you are an academic, researcher, or technical specialist in the field of sci-ences; humanities; engineering; medicine; digital technology; or the arts, it may be possible for you to apply for a Global Talent visa. This route requires you to apply to be endorsed as an internationally recognized leader or emerging leader in your particular field by an endorsing body (Arts Council Eng-land, British Academy, Royal Academy of Engineering, Royal Society, Tech Nation, UK Research and Innovation (UKRI)). However, if you are applying for a senior academic role, e.g. Professor/Reader there is an accelerated route to endorsement. Further information can be found on the UKVI website
    • Tier 2: The University is a UKVI licensed sponsor and is able to issue a Certificate of Sponsorship (CoS) to successful candidates who are offered highly skilled roles and meet the eligibility criteria. The CoS enables candidates to apply for a Tier 2 (general visa).

    Please note if you were last granted leave to stay in the UK in any Tier 2 category in the 12 months immediately preceding an application and the leave has

    • ended or expired.
    • the CoS which led to that grant of leave was issued for more than 3 months, and
    • you are either:
    • applying for entry clearance from outside the UK, or
    • you are in the UK and had a previous period of Tier 2 leave, but then changed (‘switched’) into a different immigration category and now wishes to apply again under Tier 2.

    You must wait 12 months before applying again.

    Further information about whether you require a visa and other visa routes can be found at:

    Conditions of Employment

    Pension Scheme

    This role is grade UE06 and therefore the post holder is automatically included in the membership of the Universities Superannuation Scheme (USS), subject to the USS membership criteria unless they indicate that they choose not to join the Scheme.

    For further information please visit our Pensions website.


    The role is grade UE06 and attracts an annual salary of £28,331 to £32,816 for 35 hours each week. Salary is paid monthly by direct transfer to your Bank or Building Society account, normally on the 28th of the month. Salaries for part-time staff are calculated on the full-time scales, pro-rata to the Standard Working Week.

    The University reserves the right to vary the candidate information or make no appointment at all. Neither in part nor in whole does this information form part of any contract between the University and any individual.