3. Data, Science & AI

Executes PostgreSQL queries against Supabase databases using secure 1Password credential retrieval for database access.

Retrieves and ranks framework documentation to provide accurate, source-linked answers to user queries about API usage, errors, and configuration.

Builds semantic queries with measures, dimensions, filters, and time dimensions for Drizzle Cube, enabling structured data analysis.

Performs semantic search on a RAG database to retrieve relevant content from processed documents, code, or academic papers.

Processes documents by chunking, embedding, and indexing into vector databases for RAG-based semantic search applications.

Creates and configures Drizzle Cube semantic layer definitions with security, measures, dimensions, and joins for data analytics.

Configures interactive dashboards with portlets (charts/widgets) using Drizzle Cube's DashboardConfig for data visualization and analysis.

Expertly generates optimized SQL queries for efficient database operations and data retrieval.

Creates and configures AnalysisConfig objects for query, funnel, and flow analysis in Drizzle Cube dashboards for data exploration.

Retrieves specific vector entries from Agentuity cloud platform by key, requiring authentication for secure data access in AI/ML contexts.

Systematically resolves common Scikit-learn errors including model fitting, data shape mismatches, and pipeline configuration issues.

Creates and edits XLSX spreadsheets for data analysis, budgeting, and organizational data management.

End-to-end orchestrator for processing raw workout logs into structured database records through a validated pipeline with quality gates and git tracking.

Debugs common Pandas DataFrame errors including SettingWithCopyWarning, KeyError, merge mismatches, and memory issues in Python data analysis.

Refactors Scikit-learn ML code to enhance maintainability, prevent data leakage, and ensure reproducibility through proper pipeline implementation.

Generates structured prompt files to customize AI model behavior and responses for specific applications.

Enables AI agents to persistently store and retrieve conversation history for context-aware interactions.

Systematically debugs PyTorch issues including tensor errors, CUDA memory problems, gradient issues, and shape mismatches in deep learning models.

Normalizes a single DPP passport input to the lab_battery_passport_extended_v1 format for standard data processing.

Automates creation of A/B/C experiment scaffolds including rules, inputs, and attempts in a single deterministic run.

Persists episode ingestion run data in the lab's genetic file store for data processing and analysis workflows.

Exports Google Docs and Sheets content to Markdown or stdout for seamless integration into data analysis, summarization, and AI context loading workflows.

Converts natural-language procedure descriptions into structured step lists, serving as a template for MOVA skills in the lab.

Builds deterministic evaluation suites for AI agent workflows, enabling reproducible testing with metrics, graders, and JSONL logs for capability tracking.