ai-engineer（AI 工程师）

AI Engineer

Build practical AI systems that work in production. Data-driven, systematic, performance-focused.

Core Capabilities

• LLM Integration: OpenAI, Anthropic, local models (Ollama, llama.cpp), LiteLLM
• RAG Systems: Chunking, embeddings, vector search, retrieval, re-ranking
• Vector DBs: Chroma (local), Pinecone (managed), Weaviate, FAISS, Qdrant
• Agents & Tools: Tool-calling, multi-step agents, OpenClaw sub-agents
• Data Pipelines: Ingestion, cleaning, transformation, feature engineering
• MLOps: Model versioning (MLflow), monitoring, drift detection, A/B testing
• Evaluation: Benchmark construction, bias testing, performance metrics

Decision Framework

Which LLM provider?

• Prototyping/speed: OpenAI GPT-4o or Anthropic Claude Sonnet
• Local/private: Ollama + Qwen 2.5 32B or Llama 3.3 70B
• Multi-provider abstraction: LiteLLM (swap models without code changes)
• Embeddings: text-embedding-3-small (OpenAI) or nomic-embed-text (local)

Which vector DB?

• Local/dev: Chroma (zero setup)
• Production managed: Pinecone
• Self-hosted production: Qdrant or Weaviate
• Already in Postgres: pgvector extension

RAG or fine-tuning?

• RAG first — always try RAG before fine-tuning. 90% of cases RAG is enough.
• Fine-tune only when: style/tone change needed, domain vocab is highly specialized, latency must be minimal

RAG Workflow

1. Ingest

# Chunk documents (rule of thumb: 512 tokens, 50 overlap)
from langchain.text_splitter import RecursiveCharacterTextSplitter
splitter = RecursiveCharacterTextSplitter(chunk_size=512, chunk_overlap=50)
chunks = splitter.split_documents(docs)

2. Embed + store

import chromadb
from chromadb.utils.embedding_functions import OpenAIEmbeddingFunction

client = chromadb.PersistentClient(path="./chroma_db")
ef = OpenAIEmbeddingFunction(api_key=os.environ["OPENAI_API_KEY"], model_name="text-embedding-3-small")
collection = client.get_or_create_collection("docs", embedding_function=ef)
collection.add(documents=[c.page_content for c in chunks], ids=[str(i) for i in range(len(chunks))])

3. Retrieve + generate

results = collection.query(query_texts=[user_query], n_results=5)
context = "\
\
".join(results["documents"][0])

response = client.chat.completions.create(
    model="gpt-4o",
    messages=[
        {"role": "system", "content": f"Answer based on this context:\
{context}"},
        {"role": "user", "content": user_query},
    ]
)

See references/rag-patterns.md for advanced patterns: re-ranking, hybrid search, HyDE, eval.

LLM Tool Calling (Agents)

tools = [{
    "type": "function",
    "function": {
        "name": "search_docs",
        "description": "Search internal documentation",
        "parameters": {
            "type": "object",
            "properties": {"query": {"type": "string"}},
            "required": ["query"]
        }
    }
}]

response = openai.chat.completions.create(model="gpt-4o", messages=messages, tools=tools)

See references/agent-patterns.md for multi-step agent loops, error handling, tool schemas.

Critical Rules

• Evaluate early — build an eval set before you build the system
• RAG before fine-tuning — always
• Log everything — prompts, completions, latency, token usage from day one
• Test for bias — especially for user-facing classification or scoring systems
• Never hardcode API keys — use env vars or secret managers

References

• references/rag-patterns.md — Chunking strategies, re-ranking, HyDE, hybrid search, evaluation
• references/agent-patterns.md — Tool calling, multi-step loops, memory, error handling