3. Data, Science & AI

Robust PDF processing for forms, tables, OCR, and batch operations with validation and error handling.

Manages priority, authentication, usage limits, and automatic fallback across multiple AI model providers for seamless service continuity.

Navigates Supabase database structures, relationships, and query patterns to efficiently retrieve and analyze data.

Framework for analyzing, creating, and refining AI prompts using evidence-based techniques to optimize performance for language models like Claude and ChatGPT.

Enables data quality validation implementation using Great Expectations, dbt tests, and data contracts for reliable data pipelines.

Real-time diagnostic tool for analyzing AI prompt quality, detecting anti-patterns and token waste, with actionable optimization recommendations.

Builds production RAG systems with semantic chunking, incremental indexing, and context-aware retrieval for efficient document search and generation.

Expertly crafts, optimizes, and debugs T-SQL queries for Microsoft SQL Server, including complex operations and performance tuning.

Analyzes DEX liquidity depth and estimates slippage for MEV trading, supporting optimal swap execution and route selection.

Converts PDFs and images to markdown via Claude Vision AI for accurate text extraction and documentation.

Provides AI-driven analysis of GitHub, GitLab, and Bitbucket repositories using RAG to understand codebases and answer project-specific questions.

Extracts structured data from scientific PDFs to enable systematic reviews, meta-analyses, and database creation with validation metrics.

Processes receipts via OCR, parses financial data, categorizes expenses, and integrates with databases for accurate financial extraction.

Debugs and optimizes machine learning training by addressing loss divergence, overfitting, and slow convergence issues.

Aggregates insights from multiple tasks to drive data-driven improvements and evolution in systems or processes.

Enables adaptive learning systems using AgentDB's vector database for self-learning agents, featuring trajectory tracking and memory distillation to optimize decision-making.

Designs and implements memory architectures for AI agents to persist state, maintain entity consistency, and reason over structured knowledge across sessions.

Verifies software versions, dates, status, and current state using assertions.yaml and web search to prevent AI hallucinations.

Extracts human-readable transcripts from JSONL session logs generated by Claude Code and Codex CLI tools.

Provides expert knowledge on Logseq's Datascript database schema, including attributes, relationships, and data model architecture for effective data management.

Optimizes Apache Spark jobs via partitioning, caching, shuffle tuning, and memory configuration to accelerate data processing pipelines and resolve performance bottlenecks.

Guides creation and enhancement of Claude Code agents through structured file organization, persona definition, tool integration, and schema validation.

An AI runtime toolkit for managing and integrating internal AI tools and external CLI tools with discovery, execution, and configuration capabilities.

Universal dataset importer for FiftyOne supporting all media types, label formats, and autonomous driving data with automatic structure detection.