SM
Sheby M
Loading portfolio experience
Business Operations, Marketing · AI Agent, Web App, Dashboard
SumUps AI
An AI-powered survey intelligence platform that processes large volumes of quantitative and open-text responses, discovers important themes, compares demographic groups, and produces evidence-linked summaries for policy teams.
Stack and focus areas
PythonReactAIFull StackOpenAIAnthropic ClaudeAI AgentsNatural Language ProcessingFastAPIPandasNumPyscikit-learnNext.jsTypeScriptSupabasePostgreSQLpgvectorRAGTopic ModelingSentiment AnalysisSemantic SearchClusteringSurvey AnalysisStatistical AnalysisCross-TabulationData VisualizationCSV ProcessingExcel ProcessingReport GenerationRole-Based Access ControlAWS S3DockerVercelConsultingGovernmentPublic PolicyResearchSurvey AnalyticsAnalytics DashboardData PlatformProfessional Services
Project description
Policy teams, research organizations, and public-sector analysts frequently collect thousands of survey responses containing rating questions, demographic information, multiple-choice answers, and detailed open-text feedback. Traditional analysis requires substantial manual coding and spreadsheet work. Important minority viewpoints can be missed, qualitative insights are difficult to connect with quantitative results, and producing an executive summary may take weeks after a survey closes.
I designed and developed SumUps AI as a secure survey-analysis and decision-intelligence platform. Analysts can upload CSV and Excel datasets, map survey questions, identify demographic fields, clean inconsistent values, and configure the goals of the analysis. The platform calculates quantitative statistics and uses natural-language processing to group open-text responses into themes, subthemes, concerns, suggestions, and emerging issues. AI-generated findings are linked directly to supporting responses, allowing analysts to verify every conclusion. Interactive dashboards compare results across age groups, locations, departments, service users, and other segments. A conversational analysis assistant lets policy teams ask questions about the dataset, while a report builder produces executive summaries, key findings, representative quotations, charts, and recommended areas for further investigation.
