3. Data, Science & AI

Comprehensive plan mode for improving AI model robustness and generalization through answer extraction, data validation, and multi-seed evaluation.

Plans and executes advanced AI model development workflows using cutting-edge techniques for large-scale training and architecture optimization.

Converts Git repositories into LLM-optimized text files for codebase analysis, digest creation, and repository ingestion.

Provides AI-driven mathematical theorem proving with SFT+GRPO training and agent integration for formal proof generation and scientific discovery.

Converts PDF, URL, or text documents into structured Q&A pairs for AI applications, storing results in memory.

Validates investment data accuracy by cross-referencing with authorized financial APIs and sources to ensure reliability.

Expert CLI for Jean Claude AI orchestration framework, enabling two-agent workflows, event sourcing, and real-time agent coordination monitoring.

Executes GGUF quantization with imatrix protection, performs statistical benchmarking with error bars, and generates academic documentation for AI model quantization evaluation.

Builds and executes LLM-powered data pipelines for document analysis, unstructured text processing, and ETL tasks using Anthropic Claude Haiku.

Automates stock selection by applying customizable quantitative factors such as value, growth, quality, and momentum to financial data.

Extracts structured Question-Reasoning-Answer pairs from text, validating answers against source content for domain-specific contexts.

Collects and stores investment analysis results in compliance with specified web service storage protocols.

Fetches real-time and historical stock market data including prices, fundamentals, news, and economic indicators from authorized APIs.

Extracts text transcripts from YouTube videos upon user request or URL input, providing spoken content as plain text.

Archives conversation transcripts with AI embeddings and analysis in ArangoDB for storing user stories and chain of thought.

Tracks machine learning experiments with comprehensive logging and ensures reproducibility for model training and evaluation.

Analyzes portfolio risk metrics including VaR, volatility, drawdown, concentration, and correlation, providing actionable risk management recommendations.

Enables batch LLM calls, parallel completions, and Lean4 theorem proving for mathematical proofs and formal verification.

Configures Claude API for extended reasoning, enabling complex analysis, multi-step problem solving, and deep deliberation with adjustable thinking budgets.

Generates weekly investment reports by aggregating daily data, analyzing trends, and providing strategic insights for investment decision-making.

Manages document citations, RAG patterns, and contextual retrieval for source attribution and document grounding in Claude.

Provides comprehensive fundamental, technical, and valuation analysis for individual stocks.

Optimizes AI coding agents (OpenCode/Claude) for enhanced performance in software development workflows.

Converts PDF, URL, or text content into structured Q&A pairs for AI knowledge bases, with domain-specific context support.