diff --git a/src/graphics_context.zig b/src/graphics_context.zig
index 18ee6d9..2c2cd30 100644
--- a/src/graphics_context.zig
+++ b/src/graphics_context.zig
@@ -73,9 +73,11 @@ const DeviceDispatch = vk.DeviceWrapper(.{
     .cmdEndRenderPass = true,
     .cmdBindPipeline = true,
     .cmdDraw = true,
+    .cmdDrawIndexed = true,
     .cmdSetViewport = true,
     .cmdSetScissor = true,
     .cmdBindVertexBuffers = true,
+    .cmdBindIndexBuffer = true,
     .cmdCopyBuffer = true,
 });
 
diff --git a/src/main.zig b/src/main.zig
index f17842b..cce1c18 100644
--- a/src/main.zig
+++ b/src/main.zig
@@ -8,7 +8,7 @@ const Allocator = std.mem.Allocator;
 
 const app_name = "vulkan-zig triangle example";
 
-const Vertex = struct {
+const Vertex = extern struct {
     const binding_description = vk.VertexInputBindingDescription{
         .binding = 0,
         .stride = @sizeOf(Vertex),
@@ -35,11 +35,20 @@ const Vertex = struct {
 };
 
 const vertices = [_]Vertex{
-    .{ .pos = .{ 0, -0.5 }, .color = .{ 1, 0, 0 } },
-    .{ .pos = .{ 0.5, 0.5 }, .color = .{ 0, 1, 0 } },
-    .{ .pos = .{ -0.5, 0.5 }, .color = .{ 0, 0, 1 } },
+    // .{ .pos = .{ -0.5, -0.5 }, .color = .{ 1, 0, 0 } },
+    // .{ .pos = .{ 0.5, 0.5 }, .color = .{ 0, 1, 0 } },
+    // .{ .pos = .{ -0.5, 0.5 }, .color = .{ 0, 0, 1 } },
+
+    .{ .pos = .{ -0.5, -0.5 }, .color =  .{ 1, 0, 0 } },
+    .{ .pos = .{ -0.5,  0.5 }, .color =  .{ 0, 1, 0 } },
+    .{ .pos = .{  0.5, -0.5 }, .color =  .{ 0, 0, 1 } },
+    .{ .pos = .{  0.5,  0.5 }, .color =  .{ 1, 1, 0 } },
 };
 
+const Index = u16;
+
+const indices = [_]Index{ 0, 2, 1, 1, 2, 3 };
+
 pub fn main() !void {
     if (c.glfwInit() != c.GLFW_TRUE) return error.GlfwInitFailed;
     defer c.glfwTerminate();
@@ -96,24 +105,38 @@ pub fn main() !void {
     }, null);
     defer gc.vkd.destroyCommandPool(gc.dev, pool, null);
 
-    const buffer = try gc.vkd.createBuffer(gc.dev, &.{
+    const vertex_buffer = try gc.vkd.createBuffer(gc.dev, &.{
         .size = @sizeOf(@TypeOf(vertices)),
         .usage = .{ .transfer_dst_bit = true, .vertex_buffer_bit = true },
         .sharing_mode = .exclusive,
     }, null);
-    defer gc.vkd.destroyBuffer(gc.dev, buffer, null);
-    const mem_reqs = gc.vkd.getBufferMemoryRequirements(gc.dev, buffer);
-    const memory = try gc.allocate(mem_reqs, .{ .device_local_bit = true });
-    defer gc.vkd.freeMemory(gc.dev, memory, null);
-    try gc.vkd.bindBufferMemory(gc.dev, buffer, memory, 0);
+    defer gc.vkd.destroyBuffer(gc.dev, vertex_buffer, null);
+    const vertex_mem_reqs = gc.vkd.getBufferMemoryRequirements(gc.dev, vertex_buffer);
+    const vertex_memory = try gc.allocate(vertex_mem_reqs, .{ .device_local_bit = true });
+    defer gc.vkd.freeMemory(gc.dev, vertex_memory, null);
+    try gc.vkd.bindBufferMemory(gc.dev, vertex_buffer, vertex_memory, 0);
 
-    try uploadVertices(&gc, pool, buffer);
+    try uploadData(Vertex, &gc, pool, vertex_buffer, &vertices);
+
+    const index_buffer = try gc.vkd.createBuffer(gc.dev, &.{
+        .size = @sizeOf(@TypeOf(indices)),
+        .usage = .{ .transfer_dst_bit = true, .index_buffer_bit = true },
+        .sharing_mode = .exclusive,
+    }, null);
+    defer gc.vkd.destroyBuffer(gc.dev, index_buffer, null);
+    const index_mem_reqs = gc.vkd.getBufferMemoryRequirements(gc.dev, index_buffer);
+    const index_memory = try gc.allocate(index_mem_reqs, .{ .device_local_bit = true });
+    defer gc.vkd.freeMemory(gc.dev, index_memory, null);
+    try gc.vkd.bindBufferMemory(gc.dev, index_buffer, index_memory, 0);
+
+    try uploadData(Index, &gc, pool, index_buffer, &indices);
 
     var cmdbufs = try createCommandBuffers(
         &gc,
         pool,
         allocator,
-        buffer,
+        vertex_buffer,
+        index_buffer,
         swapchain.extent,
         render_pass,
         pipeline,
@@ -152,7 +175,8 @@ pub fn main() !void {
                 &gc,
                 pool,
                 allocator,
-                buffer,
+                vertex_buffer,
+                index_buffer,
                 swapchain.extent,
                 render_pass,
                 pipeline,
@@ -167,27 +191,36 @@ pub fn main() !void {
     try gc.vkd.deviceWaitIdle(gc.dev);
 }
 
-fn uploadVertices(gc: *const GraphicsContext, pool: vk.CommandPool, buffer: vk.Buffer) !void {
+fn uploadData(comptime T: type, gc: *const GraphicsContext, pool: vk.CommandPool, buffer: vk.Buffer, source: []const T) !void {
+    // if (@typeInfo(T) == .Struct and @typeInfo(T).Struct.layout == .auto) @compileError("Requires defined T layout");
+
+    const size = @sizeOf(T) * source.len;
+
     const staging_buffer = try gc.vkd.createBuffer(gc.dev, &.{
-        .size = @sizeOf(@TypeOf(vertices)),
+        .size = size,
         .usage = .{ .transfer_src_bit = true },
         .sharing_mode = .exclusive,
     }, null);
     defer gc.vkd.destroyBuffer(gc.dev, staging_buffer, null);
+
     const mem_reqs = gc.vkd.getBufferMemoryRequirements(gc.dev, staging_buffer);
-    const staging_memory = try gc.allocate(mem_reqs, .{ .host_visible_bit = true, .host_coherent_bit = true });
+    const staging_memory = try gc.allocate(mem_reqs, .{
+        .host_visible_bit = true,
+        .host_coherent_bit = true,
+    });
     defer gc.vkd.freeMemory(gc.dev, staging_memory, null);
+
     try gc.vkd.bindBufferMemory(gc.dev, staging_buffer, staging_memory, 0);
 
     {
         const data = try gc.vkd.mapMemory(gc.dev, staging_memory, 0, vk.WHOLE_SIZE, .{});
         defer gc.vkd.unmapMemory(gc.dev, staging_memory);
 
-        const gpu_vertices: [*]Vertex = @ptrCast(@alignCast(data));
-        @memcpy(gpu_vertices, vertices[0..]);
+        const dest: [*]T = @ptrCast(@alignCast(data));
+        @memcpy(dest, source);
     }
 
-    try copyBuffer(gc, pool, buffer, staging_buffer, @sizeOf(@TypeOf(vertices)));
+    try copyBuffer(gc, pool, buffer, staging_buffer, size);
 }
 
 fn copyBuffer(gc: *const GraphicsContext, pool: vk.CommandPool, dst: vk.Buffer, src: vk.Buffer, size: vk.DeviceSize) !void {
@@ -217,6 +250,9 @@ fn copyBuffer(gc: *const GraphicsContext, pool: vk.CommandPool, dst: vk.Buffer,
         .p_command_buffers = @ptrCast(&cmdbuf),
         .p_wait_dst_stage_mask = undefined,
     };
+    // creating and submitting a queue for every copy operation seems a bad idea for "streamed" data
+    // gonna want a way to send a copy operation WITH SYNCHRONIZATION PRIMITIVES on a particular queue
+    // see https://stackoverflow.com/a/62183243
     try gc.vkd.queueSubmit(gc.graphics_queue.handle, 1, @ptrCast(&si), .null_handle);
     try gc.vkd.queueWaitIdle(gc.graphics_queue.handle);
 }
@@ -225,7 +261,8 @@ fn createCommandBuffers(
     gc: *const GraphicsContext,
     pool: vk.CommandPool,
     allocator: Allocator,
-    buffer: vk.Buffer,
+    vertex_buffer: vk.Buffer,
+    index_buffer: vk.Buffer,
     extent: vk.Extent2D,
     render_pass: vk.RenderPass,
     pipeline: vk.Pipeline,
@@ -281,8 +318,10 @@ fn createCommandBuffers(
 
         gc.vkd.cmdBindPipeline(cmdbuf, .graphics, pipeline);
         const offset = [_]vk.DeviceSize{0};
-        gc.vkd.cmdBindVertexBuffers(cmdbuf, 0, 1, @ptrCast(&buffer), &offset);
-        gc.vkd.cmdDraw(cmdbuf, vertices.len, 1, 0, 0);
+        gc.vkd.cmdBindVertexBuffers(cmdbuf, 0, 1, @ptrCast(&vertex_buffer), &offset);
+        gc.vkd.cmdBindIndexBuffer(cmdbuf, index_buffer, 0, .uint16);
+        gc.vkd.cmdDrawIndexed(cmdbuf, indices.len, 1, 0, 0, 0);
+        // gc.vkd.cmdDraw(cmdbuf, vertices.len, 1, 0, 0);
 
         gc.vkd.cmdEndRenderPass(cmdbuf);
         try gc.vkd.endCommandBuffer(cmdbuf);