Position description
Amnesty International is seeking a highly experienced consultant to undertake an external technical analysis of a major social media platform's recommender system.
The primary objectives of this consultancy are to:
- Analyze open-source code to understand the architecture, mechanisms, and risks associated with the platform's recommender system.
- Identify how the system prioritizes and amplifies content, including potential human rights risks linked to disinformation, divisive narratives, and algorithmic bias.
- Develop a methodology for auditing recommender systems to assess whether disclosed operational practices align with actual functionality.
This consultancy is part of Amnesty Tech's broader research into the role of social media platforms in amplifying harmful content and contributing to human rights risks.
Background
Amnesty International's Technology and Human Rights Programme (Amnesty Tech) is leading research efforts to scrutinize the role of digital platforms in shaping public discourse. Social media algorithms, particularly recommender systems, have been implicated in the spread of harmful content, including discriminatory narratives, extremist ideologies, and misinformation.
Following Amnesty's previous research, including the Surveillance Giants report, this consultancy will focus on analyzing publicly available technical data to assess whether engagement-driven algorithms exacerbate human rights risks.
The consultant will work closely with Amnesty Tech's Big Tech Accountability Team to provide technical expertise in evaluating recommender systems.
Objectives and Deliverables
The consultant will be responsible for:
A. Technical Analysis of Open-Source Code
- Mapping the platform's recommender system architecture, processes, and mechanisms.
- Identifying different filters, heuristics, and ranking methodologies used to prioritize and amplify content.
B. Risk Assessment
- Evaluating systemic risks such as bias amplification, echo chambers, and discriminatory content promotion.
- Assessing deviations from standard industry practices that may contribute to human rights risks.
C. Audit Methodology Development
- Designing a light-touch audit framework to verify whether disclosed policies align with observed algorithmic behaviors.
- Recommending tools and techniques to facilitate effective audits of recommender systems.
Deliverables:
- Draft Technical Report (detailed findings on the recommender system architecture, risk areas, and potential human rights impacts).
- Audit Methodology (framework for verifying disclosed policies against actual system functionality).
- Presentation of Findings (summary of technical analysis, risks, and audit recommendations).
Consultancy Duration
The consultancy will last 3 weeks (15 working days), with key milestones:
- Week 1: Initial briefing, scoping, and technical review.
- Week 1-2: Open-source code analysis.
- Week 2-3: Development of audit framework and reporting.
- Week 3-4: Final presentation and submission of deliverables.
Essential Requirements
- Proficiency in English (fluency required).
- Extensive experience in reviewing and analyzing large-scale open-source repositories, with expertise in Scala, Java, and Python.
- Strong background in recommender systems, ranking algorithms, and machine learning models.
- Demonstrated ability to conduct algorithmic audits, identifying risks such as bias, filter bubbles, and ethical concerns.
- Experience working with IDEs (e.g., JetBrains IntelliJ, Visual Studio Code) and GitHub workflows.
- Proven track record of distilling complex technical findings into actionable reports.
Desirable Qualifications
- Knowledge of human rights impacts of algorithmic systems.
- Experience with large-scale data processing and distributed systems.
- Ability to communicate technical insights to non-technical audiences.
Please refer to the Terms of Reference on our website for more information on this consultancy opportunity.
Location: Remote (global applicants welcome).
Consultancy Fee: £300 (or equivalent) per day.
Bidding Instructions
To apply for this consultancy, please upload all relevant documents to the CV section of the application portal.
- A CV highlighting relevant expertise.
- A brief technical proposal (max one page) outlining:
- Your approach to analyzing the recommender system.
- Anticipated challenges and solutions.
- A sample of previous work related to algorithmic analysis or audits.
As standard, payments are made by Amnesty International 30 days after receipt of a valid invoice for the agreed work milestone completed.
Applications must be in PDF, Word, PowerPoint or Excel format.
WHAT WE HOPE YOU WILL DO NEXT
If you are talented, passionate about human rights and want to use your skills, knowledge and experience to change the world then we would encourage you to click ‘Apply for this Role' below.
Freedom, Justice, Equality. Let's get to work.
N.B. We reserve the right to close a vacancy BEFORE the closing date in the event of an overwhelming response or a change in business priorities, therefore we encourage you to apply now if you are interested.
Application instructions
Please be sure to indicate you saw this position on Globaljobs.org
