3. Data, Science & AI

Manages local dbt-core development including installation, configuration, project setup, and workflow automation with profiles.yml and dbt_project.yml templates.

Optimizes Claude's context window by identifying essential content to retain and fetching non-essential parts on-demand, reducing token usage and improving response speed.

Queries token balances, wallet contents, and portfolio values across multiple blockchains using the Bankr SDK.

Extracts structured news content from WeChat Official Accounts, Toutiao, and major news platforms, outputting JSON or Markdown.

Quick reference guide for LangChain 1.0 core concepts including Agent, Tool, Memory, and Middleware to build AI agents and integrate LangChain.

Removes saved SourceAtlas analysis results to manage workspace and reset data state.

Lists previously saved SourceAtlas analysis results for code quality and security assessments.

Comprehensive QA harness for agentic AI systems, featuring scenario suites, determinism controls, and multi-dimensional scoring for success, safety, latency, and cost.

Systematically extracts pain points, feature gaps, and opportunities from B2B reviews and forums, building taxonomies and turning insights into experiments.

End-to-end data science workflow with MLOps, drift monitoring, and production-ready processes for AI/ML projects.

Provides production-grade SQL optimization for OLTP systems across PostgreSQL, MySQL, SQL Server, Oracle, and SQLite with comprehensive performance tuning.

Enables building reliable analytics data pipelines with dbt transformations, dimensional models, and semantic metrics for BI readiness.

Provides operational prompt engineering patterns, templates, and validation flows to optimize AI-assisted coding with Claude Code.

Engineering production-ready LLMs with strategy selection (prompting, RAG, fine-tuning), dataset design, evaluation, and deployment with cost/safety controls.

Comprehensive MLOps platform for production ML lifecycle: data ingestion, deployment, monitoring, drift detection, and security (RAG, privacy).

Enables creation, editing, and analysis of Excel spreadsheets with formulas, charts, pivot tables, and data validation using Python and Node.js libraries.

End-to-end RAG system engineering covering chunking, hybrid retrieval, reranking, and evaluation for semantic search and generation.

Provides universal patterns and tooling for building data lakes and lakehouses across ingestion, transformation, storage, query, streaming, orchestration, and visualization.

Optimizes LLM inference operations for latency, cost, and observability in production environments through advanced serving patterns.

Configures Claude Code agents via YAML frontmatter, tool selection, and model specs for specialized AI subagents handling multi-step tasks.

Enables production deployment of AI agents with agentic RAG, handoff orchestration, and multi-layer guardrails using modern best practices.

Provides operational patterns, templates, and decision rules for time series forecasting using modern ML and AI methods including tree-based, deep learning, and generative models.

Guides the design and optimization of agent tools, emphasizing complexity reduction for effective AI agent operation.

End-to-end pipeline for creating book-based SFT datasets and training style-transfer models using LoRA, including text extraction and synthetic instruction generation.