3. Data, Science & AI

Diagnoses and corrects unexpected behavior in Claude AI by analyzing and adjusting configuration files in user or project directories.

Architects database schemas for Tourly collections in Appwrite, defining attributes, indexes, and enums.

Enables autonomous AI development loops using the Ralph Wiggum technique for iterative project building and problem-solving.

Provides optimized Postgres query writing, schema design, and configuration best practices for improved database performance and reliability.

Provides clear explanations of a project's database schema, data models, and data patterns for improved data understanding and development.

Ensures correct and complete metadata for data assets, enhancing data quality and interoperability.

Builds aggregation services that calculate Performance and Potential axes from evidence data for data-driven analytics and decision support.

Conducts comprehensive database audits to identify anomalies, data inconsistencies, and referential integrity issues for quality improvement and migration readiness.

Optimizes SQL queries for Databricks analytics by analyzing table metadata, partitions, and data size to enhance performance on large datasets.

Automates shifting schedule dates after a pivot date by a specified delta in JSON or HTML format.

Generates publication-quality research figures for AI/ML papers using Matplotlib, Seaborn, and Plotly, optimizing for clarity and scientific accuracy.

Retrieves and synthesizes context from source code repositories or local paths to answer deep, non-obvious questions using LLM-based RAG.

Monitors real-time status of deep learning models and data processing pipelines for operational health checks.

Expert guidance on ClickHouse query optimization, data engineering patterns, and high-performance analytics for analytical workloads.

Activates maximum performance mode through parallel orchestration of AI agents for efficient complex task execution.

Provides rule-based sea transit Go/No-Go decisions using wave height, wind speed, squall buffers, and weather window thresholds.

Optimizes LLM context window usage by analyzing limits, loading order, and compaction behavior to enhance efficiency and reduce token costs.

Provides rapid deep learning model development and deployment utilities within the deep-loop framework.

Initiates the Ralph AI service within the deep-loop plugin for machine learning and LLM-based applications.

Automates data cleaning and preprocessing tasks for data science and AI model preparation.

Creates objective.json configurations to enable autonomous iteration toward measurable AI/ML targets such as accuracy, latency, and coverage.

Recommends optimal subagent combinations for task distribution, enabling coordinated execution across specialized AI capabilities.

Provides Redis performance optimization guidelines for data structures, key design, caching, and AI patterns including RAG and vector search.

Validates CV data against ATS-compliant schema, ensuring field constraints and 2-page layout compatibility before PDF generation.