3. Data, Science & AI

Generates, edits, and transforms images using Gemini Nano AI with aspect ratio, resolution, and image-to-image capabilities via Python scripts.

Orchestrates specialized AI agent teams to decompose requests, design minimal agent chains, and auto-execute workflows to final output.

Specializes in database schema design, normalization, indexing, and relationship definition. Creates ER diagrams and migration scripts for data modeling.

Designs A/B tests, validates hypotheses, calculates statistical significance, and generates experiment reports for data-driven decisions.

Designs KPIs, tracking events, and dashboards with GA4/Amplitude/Mixpanel integration for data-driven decision-making.

Provides fast and cost-effective search using Anthropic's Haiku AI model for efficient information retrieval.

Provides deterministic subagent selection for AI systems, ensuring reproducibility and auditability in agent activation decisions.

Analyzes financial data within the OverTime application, delivering statistical insights and visualizations for performance tracking.

Extracts market data (gold prices, gasoline, exchange rates) from Vietnamese websites for analysis and reporting.

Automates ETL pipelines, manages data warehousing, and enables analytics through scalable data processing workflows.

Enables natural language querying across multiple databases with optimized performance and visual result presentation.

Overrides default AI model (Opus) to Sonnet for Plan, code-reviewer, code-simplifier, and Resolve Conflicts agents upon explicit user instruction.

Refines user prompts via structured interviews to clarify context, constraints, and output requirements for improved LLM performance.

Converts video files into image storyboards for analysis, species identification, and behavior tracking.

Executes a full data validation workflow: fetch, analyze, extract, register, and validate data sources.

Enables structured reasoning and analysis via JU-DO-KON! RAG vector database for planning and investigation tasks.

Validates database integrity and referential consistency to ensure data correctness after data addition or before commit operations.

Displays real-time statistics for the Reality Check database, including counts of claims, sources, chains, and predictions.

Integrates Oura Ring data via API to analyze sleep, readiness, activity, and HRV, generating automated reports and productivity correlations.

Builds AI agents with Google ADK Python for multi-agent systems, workflow agents, and Vertex AI deployment.

Analyzes images, audio, and video via Gemini API, generates images/videos using Imagen 4/Veo 3, supporting vision analysis, transcription, OCR, and design extraction.

Optimizes token usage and debugs context-related issues in AI models, ensuring efficient operation within token limits and rate constraints.

Enables creation, editing, and analysis of spreadsheets including formulas, pivot tables, charts, and data visualization for data processing.

Repomix packages code repositories into AI-friendly formats (XML, Markdown, plain text) for LLM context, security audits, and third-party library analysis.