3. Data, Science & AI

Builds AI agents using Argentic, a Python microframework for async MQTT-based agents with multi-LLM support and custom tool integration.

Provides token-efficient MySQL/PostgreSQL database operations via mycli and native CLI, replacing Artisan Tinker with 87% token savings.

Efficiently extracts specific fields from JSON files using jq, avoiding full file reads to save context and improve performance for large datasets.

Framework for building multi-agent AI systems with AgentOS runtime and MCP server integration for agentic applications.

Framework for building stateful AI agents with long-term memory, enabling memory management and multi-agent system coordination.

Builds custom AI agents for Claude Code using YAML configuration, system prompts, and tool restrictions for task delegation and agent optimization.

Detects AI-generated text using vocabulary patterns, structural analysis, and model-specific fingerprints to prevent false positives.

Provides structured guidelines for SQL file management, query development, migrations, and SQLFluff linting to ensure database code quality.

Retrieves and displays message history from Sinch Conversation API with filtering by contact, conversation, channel, and time range.

Optimizes AI prompts using Genetic-Pareto Evolution methodology to improve accuracy on test cases via Pareto frontier selection and mutation techniques.

Analyzes email attachments by extracting content, classifying importance, and performing content analysis to enhance email thread understanding.

Assists in writing idiomatic MLX code for Apple Silicon, covering neural networks, array operations, and performance tuning.

Converts JSON specifications into structured markdown with AI-generated insights, visualizations, and progressive disclosure for improved data comprehension.

Generates custom, query, and script reports for Frappe applications to enable data analysis and reporting features.

Generates Frappe fixture files for test data and master data setup, streamlining data preparation for new sites and testing environments.

Assists in developing, training, evaluating, and deploying machine learning models using MLOps practices.

Generates optimized video prompts for Google's Veo 3.1 model by analyzing images and applying its five-part formula and best practices.

Optimizes SQL queries through index strategies and EXPLAIN analysis, eliminating slow queries and enhancing database performance for application efficiency.

Automates end-to-end MLOps workflows including data preparation, model training, validation, and production deployment.

Master advanced prompt engineering techniques to maximize LLM performance, reliability, and control in production environments, including output optimization and template design.

Builds RAG systems using vector databases and semantic search to enhance LLM applications with external knowledge, enabling document Q&A and knowledge integration.

Designs LLM applications using LangChain's agent, memory, and tool integration patterns for building AI agents and complex workflows.

Provides comprehensive LLM evaluation using automated metrics, human feedback, and benchmarks to measure AI application quality and performance.

Manages AI agent and skill creation, configuration, and testing for Claude Code with tool/model selection and YAML frontmatter support.