3. Data, Science & AI

Fine-tunes Text-to-Speech models using Unsloth for voice cloning and synthetic speech generation.

Plans metrics and monitoring for AI agent systems using a tiered measurement strategy to ensure effective evaluation and optimization.

Reviews database schemas for normalization, constraints, indexing, and data types to ensure optimal design and performance.

Traces AI agent decision-making, tool selection, and reasoning chains for model analysis and debugging.

Enables monitoring and management of RAG retrieval, memory operations, and context handling in AI systems.

Monitors token usage and associated costs across AI agents to optimize budget allocation and prevent overspending.

Enables semantic knowledge storage and retrieval using embeddings and vector search for meaning-based recall.

Retrieves and inspects document chunks from Qdrant vector database for debugging, content verification, and indexing investigation in AI/ML applications.

Simulates Bitcoin trading strategies with technical analysis, backtesting, Monte Carlo projections, and Sharpe ratio metrics using market APIs.

Reviews machine learning and deep learning code for PyTorch/TensorFlow, covering model architecture, training patterns, data pipelines, GPU optimization, and experiment tracking.

Builds RAG agents with Google ADK and Vertex AI for document Q&A, knowledge base search, and citation-backed responses.

Updates a single PDF in Qdrant by deleting old vector chunks and re-uploading corrected content without affecting the entire collection.

Guides configuration, debugging, and migration for PyTorch CUDA environments, ensuring compatibility with CUDA 13 toolkit and drivers.

Offers PyTorch-native NLP utilities for tokenization, vocabulary management, and DataPipe-based text dataset pipelines.

Provides foundational PyTorch operations including tensor manipulation, automatic differentiation, and model training loop setup for machine learning development.

Enables computational knowledge queries, math solving, and scientific calculations through Wolfram Alpha integration for AI agents and data applications.

Exports PDF document chunks with contextual embeddings from Qdrant vector database to plain text for analysis and inspection.

Enables robust statistical and mathematical operations on numpy arrays by excluding missing or invalid data through masking.

Automates AI agent context setup with project analysis, preset recommendations, and MCP server configuration.

Expert SQL query writing, optimization, and schema design for PostgreSQL, MySQL, SQLite, and SQL Server, including complex queries, performance tuning, and migration.

Provides PyTorch model serving with MAR packaging, custom inference handlers, and multi-GPU scaling management.

Exports fine-tuned models to GGUF format for efficient deployment in llama.cpp, Ollama, and local LLM serving environments.

Provides SQL query optimization, index strategy guidance, and EXPLAIN analysis to enhance database performance and eliminate slow queries.

Orchestrates multi-agent workflows with defined handoffs and hierarchical agent relationships for AI system coordination.