Reword some content

Author: nkanu17

python-recipes/context-engineering/notebooks/section-2-retrieved-context-engineering/02_crafting_and_optimizing_context.ipynb

@@ -9,9 +9,9 @@
     "\n",
     "# Crafting and Optimizing Context\n",
     "\n",
-    "## From RAG Basics to Production-Ready Context Engineering\n",
+    "## From RAG Basics to Practical Context Engineering\n",
     "\n",
-    "In the previous notebook, you built a working RAG system and saw why context quality matters. Now you'll learn to engineer context with production-level rigor.\n",
+    "In the previous notebook, you built a working RAG system and saw why context quality matters. Now you'll learn to engineer context with professional-level rigor.\n",
     "\n",
     "**What makes context \"good\"?**\n",
     "\n",
@@ -34,10 +34,10 @@
     "- Different chunking strategies and their trade-offs\n",
     "- How to choose based on YOUR data characteristics\n",
     "\n",
-    "**Production Pipelines:**\n",
+    "**Context Preparation Pipelines:**\n",
     "- Three pipeline architectures (Request-Time, Batch, Event-Driven)\n",
     "- How to choose based on YOUR constraints\n",
-    "- Building production-ready context preparation workflows\n",
+    "- Building reusable context preparation workflows\n",
     "\n",
     "**Time to complete:** 90-105 minutes\n",
     "\n",
@@ -540,10 +540,10 @@
     "\n",
     "| Approach | Description | Token Usage | Response Quality | Maintenance | Verdict |\n",
     "|----------|-------------|-------------|------------------|-------------|---------|\n",
-    "| **Naive** | Include all raw data | 50K tokens | Poor (generic) | Easy | ❌ Not production-ready |\n",
+    "| **Naive** | Include all raw data | 50K tokens | Poor (generic) | Easy | ❌ Not practical |\n",
     "| **RAG** | Semantic search for relevant courses | 3K tokens | Good (relevant) | Moderate | ✅ Good for most cases |\n",
-    "| **Structured Views** | Pre-compute LLM-optimized summaries | 2K tokens | Excellent (overview + details) | Higher | ✅ Best for production |\n",
-    "| **Hybrid** | Structured view + RAG | 5K tokens | Excellent (best of both) | Higher | ✅ Best for production |\n",
+    "| **Structured Views** | Pre-compute LLM-optimized summaries | 2K tokens | Excellent (overview + details) | Higher | ✅ Best for real-world use |\n",
+    "| **Hybrid** | Structured view + RAG | 5K tokens | Excellent (best of both) | Higher | ✅ Best for real-world use |\n",
     "\n",
     "Let's implement each approach and compare them."
    ]
@@ -984,7 +984,7 @@
     "**Option 3: Hybrid**\n",
     "- Combine both approaches\n",
     "- Pre-compute catalog view + RAG for details\n",
-    "- Good for: Production systems\n",
+    "- Good for: Real-world systems\n",
     "\n",
     "Let's implement all three and compare."
    ]
@@ -1646,7 +1646,7 @@
     "| **Response Quality** | ✅ Good (relevant) | ✅ Good (overview) | ✅✅ Excellent (both) |\n",
     "| **Latency** | ⚠️ Moderate (search) | ✅✅ Fast (cached) | ⚠️ Moderate (search) |\n",
     "| **Maintenance** | ✅ Low (auto-updates) | ⚠️ Higher (rebuild views) | ⚠️ Higher (both) |\n",
-    "| **Best For** | Specific queries | Overview queries | Production systems |\n",
+    "| **Best For** | Specific queries | Overview queries | Real-world systems |\n",
     "\n",
     "**Decision Process:**\n",
     "\n",
@@ -1890,7 +1890,7 @@
     "\n",
     "Table 1 shows the performance comparison across different HNSW configurations. As M increases from 16 to 64,\n",
     "we observe significant improvements in recall (0.89 to 0.97) but at the cost of increased latency (2.1ms to 8.7ms)\n",
-    "and memory usage (1.2GB to 3.8GB). The sweet spot for most production workloads is M=32 with ef_construction=200,\n",
+    "and memory usage (1.2GB to 3.8GB). The sweet spot for most real-world workloads is M=32 with ef_construction=200,\n",
     "which achieves 0.94 recall with 4.3ms latency.\n",
     "\n",
     "Table 1: HNSW Performance Comparison\n",
@@ -1921,7 +1921,7 @@
     "\n",
     "## 4. Implementation Recommendations\n",
     "\n",
-    "Based on our findings, we recommend the following configuration for production deployments:\n",
+    "Based on our findings, we recommend the following configuration for real-world deployments:\n",
     "\n",
     "```python\n",
     "# Optimal HNSW configuration for balanced performance\n",
@@ -2025,7 +2025,7 @@
     "```\n",
     "\n",
     "**Best practice:** Chunk code WITH its context and rationale\n",
-    "- ✅ \"For production deployment, we recommend M=32 and ef_construction=200 because...\"\n",
+    "- ✅ \"For real-world deployment, we recommend M=32 and ef_construction=200 because...\"\n",
     "- ❌ Don't chunk code without explaining WHY these values\n",
     "\n",
     "**3. Query-Specific Retrieval Patterns**\n",
@@ -2067,12 +2067,12 @@
     "\n",
     "There's no single \"best\" chunking strategy - the optimal approach depends on YOUR data characteristics and query patterns. Let's explore different strategies and their trade-offs.\n",
     "\n",
-    "**🔧 Using LangChain for Production-Ready Chunking**\n",
+    "**🔧 Using LangChain for Professional-Grade Chunking**\n",
     "\n",
-    "In this section, we'll use **LangChain's text splitting utilities** for Strategies 2 and 3. LangChain provides battle-tested, production-ready implementations that handle edge cases and optimize for LLM consumption.\n",
+    "In this section, we'll use **LangChain's text splitting utilities** for Strategies 2 and 3. LangChain provides battle-tested, robust implementations that handle edge cases and optimize for LLM consumption.\n",
     "\n",
     "**Why LangChain?**\n",
-    "- **Industry-standard**: Used by thousands of production applications\n",
+    "- **Industry-standard**: Used by thousands of real-world applications\n",
     "- **Smart boundary detection**: Respects natural text boundaries (paragraphs, sentences, words)\n",
     "- **Local embeddings**: Free semantic chunking with HuggingFace models (no API costs)\n",
     "- **Well-tested**: Handles edge cases (empty chunks, unicode, special characters)\n",
@@ -2917,13 +2917,13 @@
    "source": [
     "---\n",
     "\n",
-    "## Part 5: Building Production-Ready Context Pipelines\n",
+    "## Part 5: Building Practical Context Pipelines\n",
     "\n",
-    "Now that you understand data transformation and chunking, let's discuss how to build production-ready pipelines.\n",
+    "Now that you understand data transformation and chunking, let's discuss how to build reusable pipelines.\n",
     "\n",
     "### Three Pipeline Architectures\n",
     "\n",
-    "There are three main approaches to context preparation in production:\n",
+    "There are three main approaches to context preparation in real-world applications:\n",
     "\n",
     "### Architecture 1: Request-Time Processing\n",
     "\n",
@@ -3295,15 +3295,15 @@
     "**3. Three Engineering Approaches**\n",
     "- **RAG:** Semantic search for relevant data (good for specific queries)\n",
     "- **Structured Views:** Pre-computed summaries (excellent for overviews)\n",
-    "- **Hybrid:** Combine both (best for production)\n",
+    "- **Hybrid:** Combine both (best for real-world use)\n",
     "\n",
     "**4. Chunking is an Engineering Decision**\n",
     "- **Don't chunk** if data is already small and complete (< 500 tokens)\n",
     "- **Do chunk** if documents are long (> 1000 tokens) or multi-topic\n",
     "- Four strategies: Document-Based, Fixed-Size, Semantic, Hierarchical\n",
     "- Choose based on YOUR data characteristics, query patterns, and constraints\n",
     "\n",
-    "**5. Production Pipeline Architectures**\n",
+    "**5. Context Pipeline Architectures**\n",
     "- **Request-Time:** Process on-the-fly (simple, always fresh, higher latency)\n",
     "- **Batch:** Pre-process in batches (fast queries, can be stale)\n",
     "- **Event-Driven:** Process on changes (real-time, complex infrastructure)\n",
@@ -3332,7 +3332,7 @@
     "\n",
     "### The Systematic Optimization Process\n",
     "\n",
-    "Now that you understand data engineering and production pipelines, let's learn how to systematically optimize context quality.\n",
+    "Now that you understand data engineering and context pipelines, let's learn how to systematically optimize context quality.\n",
     "\n",
     "**The Process:**\n",
     "```\n",
@@ -3821,7 +3821,7 @@
     "1. **Define Domain-Specific Metrics** - Don't rely on generic benchmarks\n",
     "2. **Measure Systematically** - Baseline → Experiment → Measure → Iterate\n",
     "3. **Balance Trade-offs** - Relevance vs. Efficiency, Completeness vs. Token Budget\n",
-    "4. **Test Before Production** - Validate with real queries from your domain\n",
+    "4. **Test Before Deployment** - Validate with real queries from your domain\n",
     "5. **Iterate Continuously** - Quality optimization is ongoing, not one-time\n",
     "\n",
     "**The Engineering Mindset:**\n",
@@ -3840,11 +3840,11 @@
    "source": [
     "## 📝 Summary\n",
     "\n",
-    "You've mastered production-ready context engineering:\n",
+    "You've mastered practical context engineering:\n",
     "\n",
     "**Part 1: The Engineering Mindset**\n",
     "- ✅ Context is data requiring engineering discipline\n",
-    "- ✅ Naive approaches fail in production\n",
+    "- ✅ Naive approaches fail in real-world applications\n",
     "- ✅ Engineering mindset: Requirements → Transformation → Quality → Testing\n",
     "\n",
     "**Part 2: Data Engineering Pipeline**\n",
@@ -3863,7 +3863,7 @@
     "- ✅ Four strategies with LangChain integration\n",
     "- ✅ Trade-offs and decision criteria\n",
     "\n",
-    "**Part 5: Production Pipeline Architectures**\n",
+    "**Part 5: Context Pipeline Architectures**\n",
     "- ✅ Request-Time, Batch, Event-Driven\n",
     "- ✅ Batch processing example with data\n",
     "- ✅ Decision framework for architecture selection\n",
@@ -3873,7 +3873,7 @@
     "- ✅ Systematic optimization process\n",
     "- ✅ Baseline → Experiment → Measure → Iterate\n",
     "\n",
-    "**You're now ready to engineer production-ready context for any domain!** 🎉\n",
+    "**You're now ready to engineer practical context for any domain!** 🎉\n",
     "\n",
     "---"
    ]
@@ -3899,7 +3899,7 @@
     "- **Tool Calling:** Let the AI use functions (search, enroll, check prerequisites)\n",
     "- **LangGraph State Management:** Orchestrate complex multi-step workflows\n",
     "- **Agent Reasoning:** Plan and execute multi-step tasks\n",
-    "- **Production Patterns:** Error handling, retries, and monitoring\n",
+    "- **Practical Patterns:** Error handling, retries, and monitoring\n",
     "\n",
     "```\n",
     "Section 1: Context Engineering Fundamentals\n",