Engineering writeups and incident post-mortems for #ai.
Articles: 11In-depth engineering analysis: Model eval frameworks: why synthetic benchmarks fail to predict production agent reliability. Practical diagnostics, logs and production configuration.
In-depth engineering analysis: Designing schema-first tool interfaces for domain-specific AI subagents. Practical diagnostics, logs and production configuration.
In-depth engineering analysis: Building self-correcting agent loops with AST verification and compiler feedback. Practical diagnostics, logs and production configuration.
In-depth engineering analysis: Evaluating local code generation models: why synthetic benchmarks fail in real codebases. Practical diagnostics, logs and production configuration.
In-depth engineering analysis: Context window compression strategies for long-running AI agents. Practical diagnostics, logs and production configuration.
In-depth engineering analysis: Multi-agent orchestration: tool call streaming and token latency optimization. Practical diagnostics, logs and production configuration.
In-depth engineering analysis: Autonomous AI agents: deterministic execution loops and resilient error recovery. Practical diagnostics, logs and production configuration.
In-depth engineering analysis: Model Context Protocol (MCP): architecture of extensible AI tool servers. Practical diagnostics, logs and production configuration.
In-depth engineering analysis: Running 70B parameter models on workstation: memory bandwidth bottleneck. Practical diagnostics, logs and production configuration.
In-depth engineering analysis: Vector embeddings search in PostgreSQL: pgvector tuning and IVFFlat vs HNSW indexing. Practical diagnostics, logs and production configuration.
In-depth engineering analysis: Local LLM inference: benchmarking llama.cpp GGUF quantizations (Q4_K_M vs Q8_0). Practical diagnostics, logs and production configuration.
