Enterprise Knowledge Search
UNICEF wanted to help young people in Brazil find practical green-skills pathways and climate-action opportunities instead of leaving their sustainability interest disconnected from education and work options.
Reported impact
UNICEF received source code and a scalable assistant blueprint for connecting climate-concerned youth to green opportunities, with an initial implementation path in Brazil and room to expand to broader youth audiences.
Capgemini used the 2025 Global Data Science Challenge to deliver UNICEF a working agentic-AI blueprint for guiding young people in Brazil toward green learning and career opportunities. Teams built on curated green-job and training datasets using AWS infrastructure—Amazon Bedrock AgentCore, the Strands Agents SDK, and Amazon SageMaker AI—together with Mistral AI language models. The winning Green Career Assistant combines four layers: data wrangling to clean and map skills/jobs/training data, agentic dialogue to interpret a user's context, a knowledge graph that links roles to learning pathways in a verifiable way, and a human-in-the-loop review step for transparency and fairness. The result is an explainable recommendation assistant with token and carbon monitoring that UNICEF can implement and extend, starting with Brazil, instead of a static catalog of opportunities.
Curated green-job and training datasets are cleaned and linked into a knowledge graph, then the assistant uses agentic dialogue to interpret a user's goals and recommend relevant opportunities. AWS services and Mistral models power the reasoning layer, while human review and token/carbon monitoring keep the recommendations transparent and operationally responsible.
"AWS powered the challenge with secure, scalable cloud services and AI tools, including Amazon Bedrock AgentCore, the Strands Agents SDK, and Amazon SageMaker AI"
"Mistral AI contributed cutting-edge language models that allowed teams to build assistants capable of natural, context-rich dialogue and adaptive reasoning"
"Participants worked with curated datasets of green job listings and training programs"
"The solution integrates four key components: Data wrangling and analysis ... User understanding and interaction ... Knowledge graph construction ... Human-in-the-loop"
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
IBM Sample Aug 14, 2026 Wikimedia Deutschland Wikimedia Deutschland needed to make Wikidata’s massive, multilingual knowledge graph usable for modern generative AI applications because the existing access pattern depended on specialized query languages and complex… 180 employees IBM DataStax Astra DB · IBM watsonx.data 
NVIDIA Sample Jun 24, 2026 Techies Without Borders Healthcare workers in low- and middle-income countries needed current, evidence-based medical guidance despite limited connectivity, scarce localized content, language barriers, and heavy frontline workloads. 17,000+ healthcare practitioners supported across 19 countries CMES-AI · NVIDIA Jetson Nano Developer Kit · NVIDIA NIM 
Microsoft Sample Est. · Feb 11, 2026 YMCA Victoria YMCA Victoria needed a scalable way to give employees and volunteers across more than 150 locations consistent access to policies, procedures, and updates despite language barriers, fragmented communication channels,… 53% of those with different first languages, backgrounds, and cultures Azure AI Translator · Copilot Studio · Microsoft Cloud Infrastructure 
