3. Data, Science & AI

Optimizes AI model interactions using techniques like chain-of-thought and defensive prompting to enhance output quality and security.

Provides production-ready strategies for deploying machine learning models, covering containerization, versioning, and serving infrastructure.

Orchestrates concurrent execution of specialized AI agents for independent tasks, multi-concern reviews, and parallel exploration.

Guides scalable AI system architecture design with model routing, orchestration, and observability for production deployment.

Optimizes machine learning model serving in production using batching, caching, and model compilation to reduce latency and improve efficiency.

Constructs and refines datasets through quality checks, curation, deduplication, and synthetic generation for AI training and data quality improvement.

Optimizes AI inference through quantization, speculative decoding, and batching to reduce latency and costs.

Provides structured ML development workflows for experiment design, baseline establishment, iterative improvement, and experiment tracking.

Creates Model Context Protocol (MCP) servers to integrate custom tools and APIs with AI assistants, enabling extended capabilities for Claude.

Enables building production-ready AI applications with foundation models, including prompt engineering, RAG, agents, fine-tuning, and deployment.

Compares ML deployment strategies including batch, real-time, online/offline serving, edge, and serverless architectures.

Provides detailed information on deep neural network architectures including CNNs, RNNs, and Transformers for vision, NLP, and multimodal applications.

Strategically optimizes token usage in AI interactions to minimize costs without compromising output quality, aiding prompt design and cost management.

Builds AI agents with tool use, planning (ReAct, Plan-and-Execute), memory, and evaluation for autonomous systems and multi-step workflows.

Provides feature engineering techniques for ML systems, including extraction, transformation, selection, and feature store management.

Guides machine learning model development through selection, training pipelines, hyperparameter tuning, and evaluation strategies.

Covers core concepts for designing and implementing production-ready machine learning systems, including lifecycle management and architecture principles.

Adapts foundation models to specific domains using efficient techniques such as LoRA, QLoRA, and PEFT for cost reduction and performance enhancement.

Integrates AI/ML models for vision, audio, embeddings, and RAG implementation patterns.

Provides best practices for PyTorch, TensorFlow, and scikit-learn, covering training patterns and optimization for machine learning workflows.

Provides techniques for optimizing machine learning models, including hyperparameter tuning, architecture search, and performance profiling.

Provides expertise in PostgreSQL database design, query optimization, and best practices for efficient data management.

Provides evaluation methods for AI model outputs including exact match, semantic similarity, and LLM-as-judge to measure quality and compare models.

Fetches, scores, and ranks articles from RSS, APIs, and HTML sources to generate enhanced Chinese reports based on relevance.