Gilfeather/furnace

AI Summary: Implement a benchmarking suite for a Rust-based ML inference server using the criterion crate. The suite should measure inference time, memory usage, and throughput for varying input sizes. The benchmarks should be integrated into the CI pipeline to detect performance regressions.

AI Summary: Implement model validation during server startup for the Furnace ML inference server. This involves checking the model file's existence, readability, format, input/output dimensions, and performing a test inference. Clear error messages should be provided, and a `--skip-validation` flag should be added for advanced users. The validation should be integrated into the `src/model.rs` file, utilizing existing error types from `src/error.rs`.

AI Summary: Implement request timeout handling in a Rust-based ML inference server using the Axum web framework and Tokio runtime. This involves adding configurable timeout settings (defaulting to 30 seconds), returning appropriate HTTP 408 error codes for timeouts, updating CLI arguments, logging timeout events, writing tests, and updating documentation.

AI Summary: The task is to create comprehensive API documentation for a Rust-based ML inference server, including an OpenAPI 3.0 specification, example requests and responses, error code documentation, and HTML documentation generated from the spec. The documentation should cover all endpoints and utilize the utoipa crate for OpenAPI generation.

AI Summary: Optimize the Docker image for the Rust-based ML inference server, "Furnace", to reduce its size to under 50MB. This involves creating a multi-stage Dockerfile using a slim base image (like Alpine or distroless), copying only necessary files, potentially static linking, and updating the README with Docker instructions. A `.dockerignore` file should also be created to exclude unnecessary files during the build process.

AI Summary: Implement comprehensive integration tests for the Furnace ML inference server's HTTP API endpoints (/healthz, /model/info, /predict), including error handling and using a testing framework like reqwest. The tests should be able to start and stop the server automatically and be integrated into the CI pipeline.

AI Summary: Improve the user-friendliness and debugging capabilities of the Furnace ML inference server by enhancing error messages and implementing structured logging using the `tracing` crate. This involves replacing generic error messages with specific, actionable ones, adding structured logging for requests and responses, and including request IDs for better tracing. The `anyhow` crate should be used for improved error handling and context.

AI Summary: Implement configuration file support for the Furnace ML inference server using a TOML file. This involves creating a `Config` struct, parsing the TOML file with the `toml` and `serde` crates, integrating it into the existing CLI argument parsing with `clap`, and ensuring command-line arguments override config file settings. The configuration should include settings for port, host, model path, and log level. An example `furnace.toml` file should also be added.

AI Summary: Implement a new `/metrics` endpoint for the Rust-based ML inference server that exposes server metrics (total requests, average response time, error count) in JSON format. This involves adding metrics tracking to existing request handlers, using atomic counters for thread safety, and creating an integration test.

AI Summary: Implement input validation middleware for a Rust-based ML inference server using Axum. The middleware should check the length and value range of input arrays, returning HTTP 400 errors for invalid requests. Unit tests are required.