3. Data, Science & AI

Optimizes database performance by resolving slow queries, indexes, N+1 problems, and analyzing query execution through targeted tuning.

Optimizes token usage in AI models by compressing context windows and managing session length to stay within token limits.

Manages Claude's learned patterns, enabling listing, reviewing, archiving, and adjusting confidence (boost/penalize) for effectiveness.

Analyzes BuildBuddy billing usage and trends to identify anomalies, compare periods, and recommend cost-saving actions while securing API keys.

Automatically configures Sentry for monitoring AI agents, LLM calls, and supported SDKs like OpenAI and LangChain.

Enables context-aware conversations by integrating episodic memory for recall and knowledge graphs for persistent fact storage across sessions.

Enforces TimescaleDB 2.24.0 best practices for time-series data, including hypertables, continuous aggregates, and compression.

Configures Sentry for AI agent and LLM call monitoring, automatically integrating with OpenAI, LangChain, and other AI SDKs.

Enforces PostGIS 3.6.1 best practices for spatial queries, geometry operations, and location-based features in geographic data processing.

Comprehensive tool for designing database schemas, generating ER diagrams, SQL scripts, and documentation for MySQL, PostgreSQL, and SQL Server.

Analyzes customer support data (Excel/CSV) to categorize issues, prioritize by impact (P0-P3), and generate strategic markdown reports with trends and examples.

Generates images from text prompts using the GLM-Image API, enabling visual content creation through natural language descriptions.

Extracts meeting recordings and transcripts from Quill macOS app database, outputting structured markdown with speaker identification and metadata.

Automates AI trend briefings and technology research for job search updates on AI tools and models.

Guides users in implementing Evaluation-Driven Development (EDD) for comprehensive AI model evaluation strategies.

Expert guidance for Stan probabilistic programming development, covering modern syntax, cmdstanr/cmdstanpy integration, and testing patterns.

Guides R package development with best practices using devtools, testthat, and roxygen2 for robust, maintainable data science packages.

Extracts job descriptions and company information from LinkedIn job postings for batch import into recruitment systems.

Advises on optimal checkpointing to prevent context compaction interruptions during extended AI model operations.

Enables persistent AI model behavior control through steering vectors, allowing customized outputs and versioned model modifications.

Identifies causal pathways in neural networks by determining which components mediate specific model behaviors and information flow.

Provides core concepts for neural network interpretability, including model setup, activation tracing, and internal interventions.

Provides gradient-based approximation for activation patching to enable scalable neural circuit analysis in AI models.

Visualizes intermediate layer predictions in neural networks to understand information flow and model behavior at each layer.