AI 工程编码
Implement a basic RAG pipeline using an embedding model and a vector database.
原题Build a simple AI agent with tool use (e.g., calculator, web search).
原题Implement semantic search using embeddings and cosine similarity.
原题Write code for different text chunking strategies (fixed-size, recursive, semantic).
原题Implement a prompt template system with variable substitution.
原题Build an evaluation pipeline for LLM outputs using LLM-as-a-judge.
原题Implement streaming responses for an LLM API.
原题Build a simple vector similarity search from scratch.
原题Implement a conversation memory system for a chatbot (sliding window, summary, buffer).
原题Write code to detect and handle hallucinations in LLM outputs.
原题Implement a retry mechanism with exponential backoff for LLM API calls.
原题Write a function calling (tool use) handler for an LLM API.
原题Implement a simple re-ranker for search results.
原题Build a basic document parser that extracts text from PDFs and splits it into chunks.
原题Implement cosine similarity, dot product, and Euclidean distance functions from scratch.
原题Write code to implement token counting and context window management.
原题Build a simple prompt versioning system.
原题Implement a caching layer for LLM responses.
原题Implement semantic caching for LLM queries (cache responses for semantically similar queries).
原题Write code to detect prompt injection attempts in user inputs.
原题Implement an LLM output guardrails system that checks for off-topic responses and PII leakage.
原题Build a multi-agent system where agents have different roles and collaborate on a task.
原题Implement scaled dot-product attention with a causal mask from scratch (NumPy or PyTorch).
原题Implement multi-head attention, then convert it to grouped-query attention.
原题Implement a KV cache and single-step decode for causal multi-head attention.
原题Implement BPE (Byte Pair Encoding) training and encoding from scratch.
原题Implement top-k, top-p, and temperature sampling over a logits vector.
原题Implement an LRU cache with O(1) get/put, then add per-entry TTL.
原题Implement a token-bucket rate limiter for an LLM API where cost scales with tokens, then make it distributed.
原题Write an async batch processor that runs an LLM call over 50,000 documents with a concurrency limit, retries with jitter, and error isolation.
原题Write a streaming SSE parser for LLM token streams that handles arbitrary chunk boundaries.
原题Implement a minimal agent loop with tool dispatch, error handling, and a step budget.
原题
暂无题目