diff --git a/README.md b/README.md
index e69de29..a522f7c 100644
--- a/README.md
+++ b/README.md
@@ -0,0 +1,31 @@
+# Project Description
+
+## Running the project
+
+1. Comment out all the code in [review.cs](review.cs) file. Missing directives in that file will otherwise prevent the the program from running. And this file is not central to the project itself.
+   
+2. The easiest way to run this would be to use `dotnet run [args]`. If you want a more realistic use case, you use `dotnet publish` to create an executable binary and then execute that binary with the command-line args as normal.
+
+## Infrastructure disclaimer
+I come from a background of developing mobile applications in Dart/Flutter, with a different set of conventions (such for file naming etc). I have done some research into doing it and tried sticking to it.  But in the end, the architecture is still heavily influenced by practices in mobile development.
+
+### Structure
+
+#### Assets
+There ae 2 JSON files in the `assets` directory. These can be used as input as cli-args.
+
+#### Data modelling objects
+1. [OrderDTO](data/dtos/OrderDTO.cs): Captures the raw data received from a JSON file.
+2. [OrderModel](models/OrderModel.cs): Captures the usable part of the `OrderDTO` that will be written out to the CSV.
+
+#### Services
+1. [DemoApplicationConfiguration](services/DemoApplicationConfiguration.cs): Stores the configuration that the program ran with. Such as the command-line args passed (and hence the mode), the actual input data, and the ouput destination. Along with the file-existnece checks
+2. [CsvWriter](services/CsvWriter.cs): Writes an order to a file.
+3. Deserializing the JSON was handled by a package. Could have made this into a service ideally, for better scaling up.
+
+#### Packages used
+1. Newtonsoft.Json: For deserialzing JSON into constituent DTOs easily.
+2. CsvHelper: To write out to a CSV file.
+
+# Code review exercise
+The answer to the review exercise can be found in [Problems.md](Problems.md) file.