The International Center for Advocates Against Discrimination (ICAAD) is a human rights advocacy nonprofit equipping those most harmed by systemic inequity with tools to drive structural change. We work with governments, civil society, and marginalized communities to advance data-informed policy reforms that make justice systems more equitable, efficient, and transparent. Our global team includes human rights lawyers, social scientists, data analysts, and artivists, and we regularly engage 100+ pro bono volunteers annually across disciplines — including tech.

TrackGBV, ICAAD’s flagship program, is the first large-scale empirical analysis of judicial bias in GBV cases, manually reviewing over 6,000 cases and analyzing over 3,000 sentencing decisions in collaboration with communities across the Pacific Islands. Our TrackGBV dashboard is the only comprehensive dataset tracking judicial bias in the Pacific region capturing 20 years of case law. Our findings showed that 52% of cases were impacted by bias, leading to legal reform in Fiji and the Solomon Islands, including new judicial directives and legislative amendments that impact 6 million women and girls.

However, this work is slow and resource-intensive. Each case takes ~20 minutes to process, and Western jurisdictions will likely demand 2–3x more time due to longer judgments and greater volume. To date, the corpus of cases ICAAD and its law firm partners have manually reviewed represents over 3,000 hours of manual legal analysis and validation. To scale globally including the existing work in the Caribbean, and in future to the U.S., and India, we’re integrating Generative AI.

Project 1 – ImpartialAI – Virtual Assistant for Unbiased Judicial Decision-Making

Building on these hard-won gains, ImpartialAI will harness generative AI to automate and scale our bias-detection work in the Pacific and scale it to already existing work in the Caribbean and new jurisdictions in US and India. It will also allow us to analyze judicial decisions in real time, reducing review time, while improving accuracy and consistency.

Phase 1 – TrackGBV Automation (ongoing): Automate the extraction of structured data related to bias indicators (e.g., gender stereotyping, victim blaming) from court judgments from GBV-related case law to populate a database for analysis and chatbot applications. The automation streamlines case law review by converting judgments into machine-readable features and metrics aligned with ICAAD's TrackGBV framework.

Phase 2 – TrackGBV Chatbot (planning): Build a multi-turn, chat interface for judiciaries to query bias metrics and qualitative case insights across countries and courts on demand.

Project 2 – RightsNavigator KYR App (Upcoming)

While ImpartialAI tackles bias within judicial decision-making, our TrackGBV research revealed a deeper truth: courtroom discrimination is only one facet of a broader ecosystem of harm. Survivors often face barriers long before reaching a judge—confusing procedures, inaccessible legal information, and fragmented support systems. To address these frontline challenges, ICAAD is launching RightsNavigator: a GenAI-powered Know Your Rights (KYR) app co-designed with survivors and local partners.

Across the Caribbean—including Jamaica, Grenada, and Barbados—survivors of gender-based violence frequently lack awareness of their legal rights, face life-threatening delays in protection orders, and struggle to navigate support services. RightsNavigator will leverage GenAI and Retrieval-Augmented Generation (RAG) to deliver a mobile-first, multilingual Q&A platform that demystifies legal pathways, offers step-by-step guidance through the police, medical, and court systems, and provides dynamic, real-time referrals based on proximity and service capacity. Unlike static portals, it adapts to jurisdiction-specific legal contexts and survivor needs.

An accompanying dashboard will allow civil society organizations and court personnel to monitor survivor journeys, identify procedural bottlenecks, and improve service delivery. Volunteer roles span RAG pipeline development, vector search optimization, and mobile UX—ideal for AI professionals eager to apply their skills toward justice, safety, and empowerment.

Volunteer Opportunities

Volunteers may be engaged on one or both of the projects described above (ImpartialAI and the RightsNavigator KYR App), depending on skills and availability.

Role & Responsibilities

You will help build and evaluate the core AI components of ImpartialAI and/or RightsNavigator. Your work will center on deploying and refining LLM pipelines that extract structured data from legal texts and power real-time, user-friendly Q&A tools.

Designing and refining prompts for GPT-4o-mini and AWS Bedrock/Titan

Building and optimizing vector embeddings and RAG flow

Helping with NL-to-SQL conversions and ensuring chatbot accuracy and transparency

Setting up and monitoring ML evaluation metrics for recall, bias detection, and stability

Participating in weekly or bi-weekly check-ins to prioritize model improvements

Desirable Skills & Experience

3–5 years’ experience in ML/AI roles including strong background in prompt engineering, multi-shot classification, and ML evaluation metrics

Strong Python skills, especially with transformer models and embedding frameworks

Familiarity with VectorDBs like Pinecone, OpenSearch, or Postgres+pgvector

Proficiency with AWS SageMaker/Bedrock, Lambda, EC2, S3, KMS, and RDS/Postgres

Experience scaling ML models and pipelines across regions

Comfortable integrating ML pipelines with front-end interfaces

Time Commitment

5–10 hours/week for 9–12 weeks (options for 12–24 weeks or long-term engagement)

Location & Equipment

Fully remote; no in-person presence required

#LI-DNI