diff --git a/.github/workflows/ci.yaml b/.github/workflows/ci.yaml
index 361fcc3..9e2a5b1 100644
--- a/.github/workflows/ci.yaml
+++ b/.github/workflows/ci.yaml
@@ -2,8 +2,6 @@ name: CI
 on:
   pull_request:
   push:
-env:
-  DATABASE_URL: postgresql://georm:georm@postgres:5432/georm
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}
@@ -17,6 +15,10 @@ jobs:
       options: --security-opt seccomp=unconfined
     permissions:
       pull-requests: write
+    strategy:
+      fail-fast: false
+      matrix:
+        backend: [postgres, sqlite]
     services:
       postgres:
         image: postgres:16-alpine
@@ -29,19 +31,32 @@ jobs:
           --health-interval 10s
           --health-timeout 10s
           --health-retries 5
+    env:
+      DATABASE_URL: ${{ matrix.backend == 'postgres' && 'postgresql://georm:georm@postgres:5432/georm' || 'sqlite:///tmp/georm-ci.db' }}
     steps:
       - uses: actions/checkout@v4
       - name: Install Nix
         uses: cachix/install-nix-action@v31
       - name: Install devenv
         run: nix profile install nixpkgs#devenv
-      - name: Migrate database
-        run: devenv shell just migrate
+      - name: Create SQLite database file
+        if: matrix.backend == 'sqlite'
+        run: touch /tmp/georm-ci.db
+      - name: Migrate database (postgres)
+        if: matrix.backend == 'postgres'
+        run: devenv shell just migrate-postgres
+      - name: Migrate database (sqlite)
+        if: matrix.backend == 'sqlite'
+        run: devenv shell just migrate-sqlite
+      # Formatting and dependency audit don't vary by backend, so only run
+      # them once rather than duplicating across both matrix legs.
       - name: Formatting check
+        if: matrix.backend == 'postgres'
         run: devenv shell just format-check
-      - name: Lint
-        run: devenv shell just lint
       - name: Audit
+        if: matrix.backend == 'postgres'
         run: devenv shell just audit
+      - name: Lint
+        run: devenv shell just ${{ matrix.backend == 'postgres' && 'lint' || 'lint-sqlite' }}
       - name: Tests
-        run: devenv shell just test
+        run: devenv shell just ${{ matrix.backend == 'postgres' && 'test-postgres' || 'test-sqlite' }}
diff --git a/.gitignore b/.gitignore
index 461c3f2..1f92086 100644
--- a/.gitignore
+++ b/.gitignore
@@ -4,6 +4,12 @@
 /.sqls
 /examples/target
 
+# Local SQLite databases used for manual testing
+*.db
+*.db-journal
+*.sqlite
+*.sqlite3
+
 # Devenv
 .devenv*
 devenv.local.nix
diff --git a/Cargo.lock b/Cargo.lock
index 0131bf5..b72cc1e 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -951,6 +951,7 @@ version = "0.30.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "2e99fb7a497b1e3339bc746195567ed8d3e24945ecd636e3619d20b9de9e9149"
 dependencies = [
+ "cc",
  "pkg-config",
  "vcpkg",
 ]
diff --git a/Cargo.toml b/Cargo.toml
index 2976b62..6e3d294 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -18,7 +18,7 @@ categories = ["database"]
 [package]
 name = "georm"
 readme = "README.md"
-description = "Georm, a simple, opiniated SQLx ORM for PostgreSQL"
+description = "Georm, a simple, opiniated SQLx ORM for PostgreSQL and SQLite"
 authors.workspace = true
 edition.workspace = true
 homepage.workspace = true
@@ -26,13 +26,18 @@ license.workspace = true
 repository.workspace = true
 version.workspace = true
 
+[features]
+default = ["postgres"]
+postgres = ["georm-macros/postgres", "sqlx/postgres", "sqlx/bigdecimal"]
+sqlite = ["georm-macros/sqlite", "sqlx/sqlite"]
+
 [workspace.dependencies]
 georm-macros = { version = "=0.2.1", path = "georm-macros" }
 
 [workspace.dependencies.sqlx]
 version = "0.8.6"
 default-features = false
-features = ["postgres", "runtime-tokio", "macros", "migrate", "bigdecimal"]
+features = ["runtime-tokio", "macros", "migrate"]
 
 [dependencies]
 sqlx = { workspace = true }
@@ -43,9 +48,28 @@ chrono = { version = "0.4", features = ["serde"] }
 rand = "0.9"
 
 [dev-dependencies.sqlx]
+# Unlike the crate's own [features] above, dev-dependencies are always built
+# for `cargo test` regardless of --no-default-features/--features passed for
+# the georm package itself. Tests bind directly to sqlx::PgPool/SqlitePool,
+# so both drivers must be enabled here unconditionally.
 version = "0.8.6"
 default-features = false
-features = ["postgres", "runtime-tokio", "macros", "migrate", "chrono"]
+features = ["postgres", "sqlite", "runtime-tokio", "macros", "migrate", "chrono", "bigdecimal"]
+
+# Cargo's test auto-discovery only scans tests/*.rs (not subdirectories), so
+# the tests/sqlite/*.rs files (kept in their own directory to avoid colliding
+# with the postgres tests/models.rs module) need explicit [[test]] entries.
+[[test]]
+name = "sqlite_simple_struct"
+path = "tests/sqlite/simple_struct.rs"
+
+[[test]]
+name = "sqlite_defaultable_struct"
+path = "tests/sqlite/defaultable_struct.rs"
+
+# sqlite_composite_key intentionally deferred: UserRole's chrono::DateTime<Utc>
+# column needs per-field query_as! type overrides under SQLite (see the note
+# in tests/sqlite/models.rs) that don't exist yet.
 
 [workspace.lints.rust]
 unsafe_code = "forbid"
diff --git a/README.md b/README.md
index e52c1b2..ade9100 100644
--- a/README.md
+++ b/README.md
@@ -7,7 +7,7 @@
 <h1 align="center">Georm</h1>
 <div align="center">
  <strong>
-   A simple, type-safe SQLx ORM for PostgreSQL
+   A simple, type-safe SQLx ORM for PostgreSQL and SQLite
  </strong>
 </div>
 <br/>
@@ -606,7 +606,7 @@ Georm is designed for zero runtime overhead:
 - **Compile-time queries**: All SQL is verified at compile time
 - **No reflection**: Direct field access, no runtime introspection
 - **Minimal allocations**: Efficient use of owned vs borrowed data
-- **SQLx integration**: Leverages SQLx's optimized PostgreSQL driver
+- **SQLx integration**: Leverages SQLx's optimized PostgreSQL or SQLite driver, selected via the `postgres`/`sqlite` Cargo features
 
 ## Examples
 
@@ -651,7 +651,7 @@ cargo run help # For a list of all available actions
 
 ### High Priority
 - **Simplified Relationship Syntax**: Remove redundant table/remote_id specifications by inferring them from target entity metadata
-- **Multi-Database Support**: MySQL and SQLite support with feature flags
+- **Multi-Database Support**: SQLite support landed behind the `sqlite` feature flag (composite keys with `chrono` columns still pending); MySQL remains unstarted
 
 ### Medium Priority
 - **Composite Key Relationships**: Add relationship support (one-to-one, one-to-many, many-to-many) for entities with composite primary keys
diff --git a/georm-macros/Cargo.toml b/georm-macros/Cargo.toml
index 77c1e78..3dc68d0 100644
--- a/georm-macros/Cargo.toml
+++ b/georm-macros/Cargo.toml
@@ -10,6 +10,10 @@ repository.workspace = true
 [lib]
 proc-macro = true
 
+[features]
+postgres = []
+sqlite = []
+
 [dependencies]
 deluxe = "0.5.0"
 proc-macro2 = "1.0.93"
diff --git a/georm-macros/src/georm/defaultable_struct.rs b/georm-macros/src/georm/defaultable_struct.rs
index 135cee8..0f30618 100644
--- a/georm-macros/src/georm/defaultable_struct.rs
+++ b/georm-macros/src/georm/defaultable_struct.rs
@@ -10,6 +10,7 @@
 //! `Defaultable` trait.
 
 use crate::georm::ir::GeneratedType;
+use crate::georm::sql::{self, SqlDialect};
 
 use super::ir::{GeormField, GeormStructAttributes};
 use quote::quote;
@@ -92,11 +93,15 @@ fn generate_defaultable_trait_impl(
         });
     }
 
+    let database = sql::DIALECT.database_type();
+    let index_var = syn::Ident::new("i", proc_macro2::Span::call_site());
+    let placeholder_expr = sql::DIALECT.runtime_placeholder(&index_var);
+
     quote! {
         impl ::georm::Defaultable<#id_type, #struct_name> for #defaultable_struct_name {
             async fn create<'e, E>(&self, mut executor: E) -> ::sqlx::Result<#struct_name>
             where
-                E: ::sqlx::Executor<'e, Database = ::sqlx::Postgres>
+                E: ::sqlx::Executor<'e, Database = #database>
             {
                 let mut dynamic_fields = Vec::new();
 
@@ -106,7 +111,7 @@ fn generate_defaultable_trait_impl(
                 all_fields.extend(dynamic_fields);
 
                 let placeholders: Vec<String> = (1..=all_fields.len())
-                    .map(|i| format!("${}", i))
+                    .map(|#index_var| #placeholder_expr)
                     .collect();
 
                 let query = format!(
diff --git a/georm-macros/src/georm/sql/mod.rs b/georm-macros/src/georm/sql/mod.rs
index b8e3cbf..e44a9d0 100644
--- a/georm-macros/src/georm/sql/mod.rs
+++ b/georm-macros/src/georm/sql/mod.rs
@@ -4,10 +4,22 @@ use crate::georm::ir::GeormField;
 use quote::quote;
 
 mod postgres;
-pub use postgres::PostgresDialect;
+mod sqlite;
 
+#[cfg(feature = "postgres")]
+pub use postgres::PostgresDialect;
+#[cfg(feature = "sqlite")]
+pub use sqlite::SqliteDialect;
+
+#[cfg(feature = "postgres")]
 pub type ActiveDialect = PostgresDialect;
+#[cfg(feature = "sqlite")]
+pub type ActiveDialect = SqliteDialect;
+
+#[cfg(feature = "postgres")]
 pub const DIALECT: ActiveDialect = PostgresDialect;
+#[cfg(feature = "sqlite")]
+pub const DIALECT: ActiveDialect = SqliteDialect;
 
 /// How a relation-lookup query should fetch its result.
 pub enum FetchKind {
@@ -29,6 +41,13 @@ pub trait SqlDialect {
     fn database_type(&self) -> proc_macro2::TokenStream;
     fn row_type(&self) -> proc_macro2::TokenStream;
 
+    /// Like [`SqlDialect::placeholder`], but for placeholder lists whose
+    /// length is only known at runtime (the defaultable-struct insert, which
+    /// varies its column count based on which `Option` fields are `Some`).
+    /// `index_var` names the runtime `usize` loop variable holding the 1-based
+    /// position of the placeholder being built.
+    fn runtime_placeholder(&self, index_var: &syn::Ident) -> proc_macro2::TokenStream;
+
     fn generate_from_row(
         &self,
         struct_name: &syn::Ident,
@@ -325,7 +344,11 @@ pub trait SqlDialect {
             where
                 E: ::sqlx::Executor<'e, Database = #database>
             {
-                ::sqlx::query_as!(#entity, #query, #arg).#fetch_method(executor).await
+                // Bound to a place rather than passed inline: sqlx's SQLite
+                // query macros need the bind argument to outlive the
+                // temporary produced by expressions like `self.get_id()`.
+                let arg = #arg;
+                ::sqlx::query_as!(#entity, #query, arg).#fetch_method(executor).await
             }
         }
     }
diff --git a/georm-macros/src/georm/sql/postgres.rs b/georm-macros/src/georm/sql/postgres.rs
index 876bc90..6e2004a 100644
--- a/georm-macros/src/georm/sql/postgres.rs
+++ b/georm-macros/src/georm/sql/postgres.rs
@@ -15,4 +15,8 @@ impl SqlDialect for PostgresDialect {
     fn row_type(&self) -> proc_macro2::TokenStream {
         quote! { ::sqlx::postgres::PgRow }
     }
+
+    fn runtime_placeholder(&self, index_var: &syn::Ident) -> proc_macro2::TokenStream {
+        quote! { format!("${}", #index_var) }
+    }
 }
diff --git a/georm-macros/src/georm/sql/sqlite.rs b/georm-macros/src/georm/sql/sqlite.rs
new file mode 100644
index 0000000..c5d0f7a
--- /dev/null
+++ b/georm-macros/src/georm/sql/sqlite.rs
@@ -0,0 +1,29 @@
+use super::SqlDialect;
+use quote::quote;
+
+pub struct SqliteDialect;
+
+impl SqlDialect for SqliteDialect {
+    fn placeholder(&self, _index: usize) -> String {
+        // SQLite accepts unnumbered `?` placeholders resolved in positional
+        // (left-to-right) order, which matches the order Georm binds values in.
+        "?".to_string()
+    }
+
+    fn database_type(&self) -> proc_macro2::TokenStream {
+        quote! { ::sqlx::Sqlite }
+    }
+
+    fn row_type(&self) -> proc_macro2::TokenStream {
+        quote! { ::sqlx::sqlite::SqliteRow }
+    }
+
+    fn runtime_placeholder(&self, index_var: &syn::Ident) -> proc_macro2::TokenStream {
+        quote! {
+            {
+                let _ = #index_var;
+                "?".to_string()
+            }
+        }
+    }
+}
diff --git a/justfile b/justfile
index 85f5a3d..4e755f1 100644
--- a/justfile
+++ b/justfile
@@ -3,18 +3,36 @@ default: lint
 clean:
 	cargo clean
 
-test:
+# Runs the whole workspace (examples included) against the default
+# "postgres" feature.
+test: test-postgres
+
+test-postgres:
 	cargo test --all-targets --all
 
+# Scoped to -p georm: examples/postgres/* depends on georm's default
+# ("postgres") feature via Cargo feature unification, so it must stay out of
+# a --no-default-features build rather than be dragged into it.
+test-sqlite:
+	cargo test -p georm --no-default-features --features sqlite --all-targets
+
 lint:
 	cargo clippy --all-targets
 
+lint-sqlite:
+	cargo clippy -p georm -p georm-macros --no-default-features --features sqlite --all-targets
+
 audit:
 	cargo deny check all
 
-migrate:
+migrate: migrate-postgres
+
+migrate-postgres:
 	cargo sqlx migrate run
 
+migrate-sqlite:
+	cargo sqlx migrate run --source migrations/sqlite
+
 build:
 	cargo build
 
@@ -27,7 +45,7 @@ format:
 format-check:
 	cargo fmt --check --all
 
-check-all: format-check lint audit test
+check-all: format-check lint audit test test-sqlite
 
 ## Local Variables:
 ## mode: makefile
diff --git a/migrations/sqlite/20250126153330_simple-struct-tests.down.sql b/migrations/sqlite/20250126153330_simple-struct-tests.down.sql
new file mode 100644
index 0000000..ad6a6e3
--- /dev/null
+++ b/migrations/sqlite/20250126153330_simple-struct-tests.down.sql
@@ -0,0 +1,6 @@
+DROP TABLE IF EXISTS reviews;
+DROP TABLE IF EXISTS book_genres;
+DROP TABLE IF EXISTS books;
+DROP TABLE IF EXISTS genres;
+DROP TABLE IF EXISTS authors;
+DROP TABLE IF EXISTS biographies;
diff --git a/migrations/sqlite/20250126153330_simple-struct-tests.up.sql b/migrations/sqlite/20250126153330_simple-struct-tests.up.sql
new file mode 100644
index 0000000..56b3a64
--- /dev/null
+++ b/migrations/sqlite/20250126153330_simple-struct-tests.up.sql
@@ -0,0 +1,38 @@
+CREATE TABLE biographies (
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    content TEXT NOT NULL
+);
+
+CREATE TABLE authors (
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    name VARCHAR(100) NOT NULL,
+    biography_id INT,
+    FOREIGN KEY (biography_id) REFERENCES biographies(id)
+);
+
+CREATE TABLE books (
+    ident INTEGER PRIMARY KEY AUTOINCREMENT,
+    title VARCHAR(100) NOT NULL,
+    author_id INT NOT NULL,
+    FOREIGN KEY (author_id) REFERENCES authors(id) ON DELETE CASCADE
+);
+
+CREATE TABLE reviews (
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    book_id INT NOT NULL,
+    review TEXT NOT NULL,
+    FOREIGN KEY (book_id) REFERENCES books(ident) ON DELETE CASCADE
+);
+
+CREATE TABLE genres (
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    name VARCHAR(100) NOT NULL
+);
+
+CREATE TABLE book_genres (
+    book_id INT NOT NULL,
+    genre_id INT NOT NULL,
+    PRIMARY KEY (book_id, genre_id),
+    FOREIGN KEY (book_id) REFERENCES books(ident) ON DELETE CASCADE,
+    FOREIGN KEY (genre_id) REFERENCES genres(id) ON DELETE CASCADE
+);
diff --git a/migrations/sqlite/20250609181248_composite-key.down.sql b/migrations/sqlite/20250609181248_composite-key.down.sql
new file mode 100644
index 0000000..d0c556b
--- /dev/null
+++ b/migrations/sqlite/20250609181248_composite-key.down.sql
@@ -0,0 +1,2 @@
+-- Add down migration script here
+DROP TABLE IF EXISTS UserRoles;
diff --git a/migrations/sqlite/20250609181248_composite-key.up.sql b/migrations/sqlite/20250609181248_composite-key.up.sql
new file mode 100644
index 0000000..b9bb071
--- /dev/null
+++ b/migrations/sqlite/20250609181248_composite-key.up.sql
@@ -0,0 +1,7 @@
+-- Add up migration script here
+CREATE TABLE UserRoles (
+    user_id INTEGER NOT NULL,
+    role_id INTEGER NOT NULL,
+    assigned_at TEXT NOT NULL DEFAULT (strftime('%Y-%m-%dT%H:%M:%fZ', 'now')),
+    PRIMARY KEY (user_id, role_id)
+);
diff --git a/src/database.rs b/src/database.rs
new file mode 100644
index 0000000..3a3bdee
--- /dev/null
+++ b/src/database.rs
@@ -0,0 +1,7 @@
+//! The SQL backend selected via Cargo feature flags (`postgres` or `sqlite`,
+//! mutually exclusive — see the guard in `lib.rs`).
+
+#[cfg(feature = "postgres")]
+pub type ActiveDatabase = ::sqlx::Postgres;
+#[cfg(feature = "sqlite")]
+pub type ActiveDatabase = ::sqlx::Sqlite;
diff --git a/src/defaultable.rs b/src/defaultable.rs
index 4b9bff9..3935c8f 100644
--- a/src/defaultable.rs
+++ b/src/defaultable.rs
@@ -1,4 +1,5 @@
-use sqlx::{Executor, Postgres};
+use crate::ActiveDatabase;
+use sqlx::Executor;
 
 /// Trait for creating entities with database defaults and auto-generated values.
 ///
@@ -277,5 +278,5 @@ pub trait Defaultable<Id, Entity> {
     ) -> impl std::future::Future<Output = sqlx::Result<Entity>> + Send
     where
         Self: Sized,
-        E: Executor<'e, Database = Postgres>;
+        E: Executor<'e, Database = ActiveDatabase>;
 }
diff --git a/src/georm.rs b/src/georm.rs
index 623213f..5946fb8 100644
--- a/src/georm.rs
+++ b/src/georm.rs
@@ -1,4 +1,5 @@
-use sqlx::{Executor, Postgres};
+use crate::ActiveDatabase;
+use sqlx::Executor;
 
 /// Core database operations trait for Georm entities.
 ///
@@ -125,7 +126,7 @@ pub trait Georm<Id> {
     ) -> impl ::std::future::Future<Output = ::sqlx::Result<Vec<Self>>> + Send
     where
         Self: Sized,
-        E: Executor<'e, Database = Postgres>;
+        E: Executor<'e, Database = ActiveDatabase>;
 
     /// Find a single entity by its primary key.
     ///
@@ -161,7 +162,7 @@ pub trait Georm<Id> {
     ) -> impl std::future::Future<Output = sqlx::Result<Option<Self>>> + Send
     where
         Self: Sized,
-        E: Executor<'e, Database = Postgres>;
+        E: Executor<'e, Database = ActiveDatabase>;
 
     /// Insert this entity as a new record in the database.
     ///
@@ -201,7 +202,7 @@ pub trait Georm<Id> {
     ) -> impl std::future::Future<Output = sqlx::Result<Self>> + Send
     where
         Self: Sized,
-        E: Executor<'e, Database = Postgres>;
+        E: Executor<'e, Database = ActiveDatabase>;
 
     /// Update an existing entity in the database.
     ///
@@ -240,7 +241,7 @@ pub trait Georm<Id> {
     ) -> impl std::future::Future<Output = sqlx::Result<Self>> + Send
     where
         Self: Sized,
-        E: Executor<'e, Database = Postgres>;
+        E: Executor<'e, Database = ActiveDatabase>;
 
     /// Insert or update this entity using PostgreSQL's upsert functionality.
     ///
@@ -276,7 +277,7 @@ pub trait Georm<Id> {
     fn upsert<'e, E>(&self, executor: E) -> impl ::std::future::Future<Output = sqlx::Result<Self>>
     where
         Self: Sized,
-        E: Executor<'e, Database = Postgres>;
+        E: Executor<'e, Database = ActiveDatabase>;
 
     #[deprecated(since = "0.3.0", note = "Please use `upsert` instead")]
     fn create_or_update<'e, E>(
@@ -285,7 +286,7 @@ pub trait Georm<Id> {
     ) -> impl ::std::future::Future<Output = sqlx::Result<Self>>
     where
         Self: Sized,
-        E: Executor<'e, Database = Postgres>,
+        E: Executor<'e, Database = ActiveDatabase>,
     {
         self.upsert(executor)
     }
@@ -324,7 +325,7 @@ pub trait Georm<Id> {
         executor: E,
     ) -> impl std::future::Future<Output = sqlx::Result<u64>> + Send
     where
-        E: Executor<'e, Database = Postgres>;
+        E: Executor<'e, Database = ActiveDatabase>;
 
     /// Delete an entity by its primary key without needing an entity instance.
     ///
@@ -364,7 +365,7 @@ pub trait Georm<Id> {
         id: &Id,
     ) -> impl std::future::Future<Output = sqlx::Result<u64>> + Send
     where
-        E: Executor<'e, Database = Postgres>;
+        E: Executor<'e, Database = ActiveDatabase>;
 
     /// Get the primary key of this entity.
     ///
diff --git a/src/lib.rs b/src/lib.rs
index 186a437..b60dc42 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,6 +1,7 @@
 //! # Georm
 //!
-//! A simple, type-safe PostgreSQL ORM built on SQLx with zero runtime overhead.
+//! A simple, type-safe ORM for PostgreSQL and SQLite built on SQLx with zero
+//! runtime overhead.
 //!
 //! ## Quick Start
 //!
@@ -618,8 +619,16 @@
 //!
 //! ### Database Support
 //!
-//! Georm is currently limited to PostgreSQL. Other databases may be supported in
-//! the future, such as SQLite or MySQL, but that is not the case yet.
+//! Georm supports PostgreSQL and SQLite, selected via the mutually exclusive
+//! `postgres` (default) and `sqlite` Cargo features. Basic CRUD (`find_all`,
+//! `find`, `create`, `update`, `upsert`, `delete`) and `Defaultable` are
+//! supported on both backends; relationship queries (`one_to_one`,
+//! `one_to_many`, `many_to_many`, field-level `relation`) are implemented for
+//! both dialects at the SQL-generation level, but composite-key entities with
+//! `chrono::DateTime` columns are not yet covered by the SQLite test suite
+//! (SQLite's compile-time query macros need explicit column type overrides
+//! for non-primitive types, which is follow-up work). `sqlx::types::BigDecimal`
+//! fields require the `postgres` feature — SQLite has no equivalent type.
 //!
 //! ### Identifiers
 //!
@@ -635,7 +644,8 @@
 //! - **No advanced queries**: No complex WHERE clauses or joins beyond relationships
 //! - **No eager loading**: Each relationship call is a separate database query
 //! - **No field-based queries**: No `find_by_{field_name}` methods generated automatically
-//! - **PostgreSQL only**: No support for other database systems
+//! - **SQLite + `chrono`**: composite-key entities with `chrono::DateTime` columns aren't yet
+//!   covered under the `sqlite` feature (see "Database Support" above)
 //!
 //! ## Generated Code
 //!
@@ -646,8 +656,15 @@
 //! - Relationship methods for accessing related entities
 //! - All CRUD operations with proper PostgreSQL optimizations
 
+#[cfg(all(feature = "postgres", feature = "sqlite"))]
+compile_error!("georm: \"postgres\" and \"sqlite\" are mutually exclusive; enable only one");
+#[cfg(not(any(feature = "postgres", feature = "sqlite")))]
+compile_error!("georm: enable exactly one of \"postgres\" or \"sqlite\"");
+
 pub use georm_macros::Georm;
 
+mod database;
+pub use database::ActiveDatabase;
 mod georm;
 pub use georm::Georm;
 mod defaultable;
diff --git a/tests/composite_key.rs b/tests/composite_key.rs
index 7ef1ba1..5d9733d 100644
--- a/tests/composite_key.rs
+++ b/tests/composite_key.rs
@@ -1,3 +1,5 @@
+#![cfg(feature = "postgres")]
+
 use georm::Georm;
 
 mod models;
diff --git a/tests/defaultable_struct.rs b/tests/defaultable_struct.rs
index 8585f57..98ed4dd 100644
--- a/tests/defaultable_struct.rs
+++ b/tests/defaultable_struct.rs
@@ -1,3 +1,5 @@
+#![cfg(feature = "postgres")]
+
 use georm::Georm;
 
 // Test struct with defaultable fields using existing table structure
diff --git a/tests/generated.rs b/tests/generated.rs
index 2dfe73c..c42e88b 100644
--- a/tests/generated.rs
+++ b/tests/generated.rs
@@ -1,3 +1,5 @@
+#![cfg(feature = "postgres")]
+
 use georm::{Defaultable, Georm};
 use sqlx::types::BigDecimal;
 
diff --git a/tests/m2m_relationship.rs b/tests/m2m_relationship.rs
index 395239c..96b0079 100644
--- a/tests/m2m_relationship.rs
+++ b/tests/m2m_relationship.rs
@@ -1,3 +1,5 @@
+#![cfg(feature = "postgres")]
+
 use georm::Georm;
 
 mod models;
diff --git a/tests/models.rs b/tests/models.rs
index 2c78e0d..94434d0 100644
--- a/tests/models.rs
+++ b/tests/models.rs
@@ -1,3 +1,5 @@
+#![cfg(feature = "postgres")]
+
 use georm::Georm;
 use sqlx::{Postgres, types::BigDecimal};
 
diff --git a/tests/o2m_relationship.rs b/tests/o2m_relationship.rs
index ffcc025..038166d 100644
--- a/tests/o2m_relationship.rs
+++ b/tests/o2m_relationship.rs
@@ -1,3 +1,5 @@
+#![cfg(feature = "postgres")]
+
 use georm::Georm;
 
 mod models;
diff --git a/tests/o2o_relationship.rs b/tests/o2o_relationship.rs
index 426c36c..59408ec 100644
--- a/tests/o2o_relationship.rs
+++ b/tests/o2o_relationship.rs
@@ -1,3 +1,5 @@
+#![cfg(feature = "postgres")]
+
 use georm::Georm;
 
 mod models;
diff --git a/tests/simple_struct.rs b/tests/simple_struct.rs
index 7ea6e1b..db7f9b6 100644
--- a/tests/simple_struct.rs
+++ b/tests/simple_struct.rs
@@ -1,3 +1,5 @@
+#![cfg(feature = "postgres")]
+
 use georm::Georm;
 use rand::seq::SliceRandom;
 
diff --git a/tests/sqlite/defaultable_struct.rs b/tests/sqlite/defaultable_struct.rs
new file mode 100644
index 0000000..44fc40a
--- /dev/null
+++ b/tests/sqlite/defaultable_struct.rs
@@ -0,0 +1,329 @@
+#![cfg(feature = "sqlite")]
+
+use georm::Georm;
+
+// Test struct with defaultable fields using existing table structure
+#[derive(Georm, Debug)]
+#[georm(table = "authors")]
+struct TestAuthor {
+    #[georm(id, defaultable)]
+    pub id: i64,
+    pub name: String,
+    pub biography_id: Option<i64>, // Don't mark Option fields as defaultable
+}
+
+// Test struct with only ID defaultable
+#[derive(Georm)]
+#[georm(table = "authors")]
+struct MinimalDefaultable {
+    #[georm(id, defaultable)]
+    pub id: i64,
+    pub name: String,
+    pub biography_id: Option<i64>,
+}
+
+// Test struct with multiple defaultable fields
+#[derive(Georm)]
+#[georm(table = "authors")]
+struct MultiDefaultable {
+    #[georm(id, defaultable)]
+    pub id: i64,
+    #[georm(defaultable)]
+    pub name: String,
+    pub biography_id: Option<i64>,
+}
+
+#[test]
+fn defaultable_struct_should_exist() {
+    let _author_default = TestAuthorDefault {
+        id: Some(1),
+        name: "Test Author".to_string(),
+        biography_id: None,
+    };
+}
+
+#[test]
+fn minimal_defaultable_struct_should_exist() {
+    let _minimal_default = MinimalDefaultableDefault {
+        id: None,
+        name: "testuser".to_string(),
+        biography_id: None,
+    };
+}
+
+#[test]
+fn defaultable_fields_can_be_none() {
+    let _author_default = TestAuthorDefault {
+        id: None,
+        name: "Test Author".to_string(),
+        biography_id: None,
+    };
+}
+
+#[test]
+fn field_visibility_is_preserved() {
+    let _author_default = TestAuthorDefault {
+        id: Some(1),
+        name: "Test".to_string(),
+        biography_id: Some(1),
+    };
+}
+
+mod defaultable_tests {
+    use super::*;
+    use georm::Defaultable;
+    use sqlx::SqlitePool;
+
+    #[sqlx::test(
+        migrations = "./migrations/sqlite",
+        fixtures(path = "fixtures", scripts("simple_struct"))
+    )]
+    async fn test_create_entity_from_defaultable_with_id(pool: SqlitePool) {
+        let author_default = TestAuthorDefault {
+            id: Some(999),
+            name: "John Doe".to_string(),
+            biography_id: None,
+        };
+
+        let created_author = author_default.create(&pool).await.unwrap();
+
+        assert_eq!(created_author.id, 999);
+        assert_eq!(created_author.name, "John Doe");
+        assert_eq!(created_author.biography_id, None);
+    }
+
+    #[sqlx::test(
+        migrations = "./migrations/sqlite",
+        fixtures(path = "fixtures", scripts("simple_struct"))
+    )]
+    async fn test_create_entity_from_defaultable_without_id(pool: SqlitePool) {
+        let author_default = TestAuthorDefault {
+            id: None,
+            name: "Jane Smith".to_string(),
+            biography_id: None,
+        };
+
+        let created_author = author_default.create(&pool).await.unwrap();
+
+        assert!(created_author.id > 0);
+        assert_eq!(created_author.name, "Jane Smith");
+        assert_eq!(created_author.biography_id, None);
+    }
+
+    #[sqlx::test(
+        migrations = "./migrations/sqlite",
+        fixtures(path = "fixtures", scripts("simple_struct"))
+    )]
+    async fn test_create_entity_from_minimal_defaultable(pool: SqlitePool) {
+        let minimal_default = MinimalDefaultableDefault {
+            id: None,
+            name: "Alice Wonder".to_string(),
+            biography_id: Some(1),
+        };
+
+        let created_author = minimal_default.create(&pool).await.unwrap();
+
+        assert!(created_author.id > 0);
+        assert_eq!(created_author.name, "Alice Wonder");
+        assert_eq!(created_author.biography_id, Some(1));
+    }
+
+    #[sqlx::test(
+        migrations = "./migrations/sqlite",
+        fixtures(path = "fixtures", scripts("simple_struct"))
+    )]
+    async fn test_create_multiple_entities_from_defaultable(pool: SqlitePool) {
+        let author1_default = TestAuthorDefault {
+            id: None,
+            name: "Author One".to_string(),
+            biography_id: None,
+        };
+
+        let author2_default = TestAuthorDefault {
+            id: None,
+            name: "Author Two".to_string(),
+            biography_id: None,
+        };
+
+        let created_author1 = author1_default.create(&pool).await.unwrap();
+        let created_author2 = author2_default.create(&pool).await.unwrap();
+
+        assert!(created_author1.id > 0);
+        assert!(created_author2.id > 0);
+        assert_ne!(created_author1.id, created_author2.id);
+
+        assert_eq!(created_author1.name, "Author One");
+        assert_eq!(created_author2.name, "Author Two");
+    }
+
+    #[sqlx::test(
+        migrations = "./migrations/sqlite",
+        fixtures(path = "fixtures", scripts("simple_struct"))
+    )]
+    async fn test_multiple_defaultable_fields_mixed(pool: SqlitePool) {
+        let multi_default = MultiDefaultableDefault {
+            id: None,
+            name: Some("Explicit Name".to_string()),
+            biography_id: Some(1),
+        };
+
+        let created = multi_default.create(&pool).await.unwrap();
+
+        assert!(created.id > 0);
+        assert_eq!(created.name, "Explicit Name");
+        assert_eq!(created.biography_id, Some(1));
+    }
+
+    #[sqlx::test(
+        migrations = "./migrations/sqlite",
+        fixtures(path = "fixtures", scripts("simple_struct"))
+    )]
+    async fn test_multiple_defaultable_fields_all_explicit(pool: SqlitePool) {
+        let multi_default = MultiDefaultableDefault {
+            id: Some(888),
+            name: Some("All Explicit".to_string()),
+            biography_id: None,
+        };
+
+        let created = multi_default.create(&pool).await.unwrap();
+
+        assert_eq!(created.id, 888);
+        assert_eq!(created.name, "All Explicit");
+        assert_eq!(created.biography_id, None);
+    }
+
+    #[sqlx::test(
+        migrations = "./migrations/sqlite",
+        fixtures(path = "fixtures", scripts("simple_struct"))
+    )]
+    async fn test_error_duplicate_id(pool: SqlitePool) {
+        let author1 = TestAuthorDefault {
+            id: Some(777),
+            name: "First Author".to_string(),
+            biography_id: None,
+        };
+
+        let author2 = TestAuthorDefault {
+            id: Some(777),
+            name: "Second Author".to_string(),
+            biography_id: None,
+        };
+
+        let _created1 = author1.create(&pool).await.unwrap();
+
+        let result2 = author2.create(&pool).await;
+        assert!(result2.is_err());
+    }
+
+    mod sql_validation_tests {
+        use super::*;
+
+        #[sqlx::test(
+            migrations = "./migrations/sqlite",
+            fixtures(path = "fixtures", scripts("simple_struct"))
+        )]
+        async fn test_sql_generation_no_defaultable_fields(pool: SqlitePool) {
+            let author_default = TestAuthorDefault {
+                id: Some(100),
+                name: "Test Name".to_string(),
+                biography_id: Some(1),
+            };
+
+            let created = author_default.create(&pool).await.unwrap();
+
+            assert_eq!(created.id, 100);
+            assert_eq!(created.name, "Test Name");
+            assert_eq!(created.biography_id, Some(1));
+
+            let found: TestAuthor = sqlx::query_as!(
+                TestAuthor,
+                "SELECT id, name, biography_id FROM authors WHERE id = ?",
+                100
+            )
+            .fetch_one(&pool)
+            .await
+            .unwrap();
+
+            assert_eq!(found.id, 100);
+            assert_eq!(found.name, "Test Name");
+            assert_eq!(found.biography_id, Some(1));
+        }
+
+        #[sqlx::test(
+            migrations = "./migrations/sqlite",
+            fixtures(path = "fixtures", scripts("simple_struct"))
+        )]
+        async fn test_sql_generation_with_defaultable_none(pool: SqlitePool) {
+            let author_default = TestAuthorDefault {
+                id: None,
+                name: "Auto ID Test".to_string(),
+                biography_id: None,
+            };
+
+            let created = author_default.create(&pool).await.unwrap();
+
+            assert!(created.id > 0);
+            assert_eq!(created.name, "Auto ID Test");
+            assert_eq!(created.biography_id, None);
+        }
+
+        #[sqlx::test(
+            migrations = "./migrations/sqlite",
+            fixtures(path = "fixtures", scripts("simple_struct"))
+        )]
+        async fn test_placeholder_ordering_consistency(pool: SqlitePool) {
+            let record2 = MultiDefaultableDefault {
+                id: Some(201),
+                name: Some("Full Record".to_string()),
+                biography_id: Some(1),
+            };
+
+            let record3 = MultiDefaultableDefault {
+                id: None,
+                name: Some("Mixed Record".to_string()),
+                biography_id: None,
+            };
+
+            let created2 = record2.create(&pool).await.unwrap();
+            assert_eq!(created2.id, 201);
+            assert_eq!(created2.name, "Full Record");
+            assert_eq!(created2.biography_id, Some(1));
+
+            let created3 = record3.create(&pool).await.unwrap();
+            assert!(created3.id > 0);
+            assert_eq!(created3.name, "Mixed Record");
+            assert_eq!(created3.biography_id, None);
+        }
+
+        #[sqlx::test(
+            migrations = "./migrations/sqlite",
+            fixtures(path = "fixtures", scripts("simple_struct"))
+        )]
+        async fn test_returning_clause_functionality(pool: SqlitePool) {
+            let author_default = TestAuthorDefault {
+                id: None,
+                name: "Return Test".to_string(),
+                biography_id: None,
+            };
+
+            let created = author_default.create(&pool).await.unwrap();
+
+            assert!(created.id > 0);
+            assert_eq!(created.name, "Return Test");
+            assert_eq!(created.biography_id, None);
+
+            let verified: TestAuthor = sqlx::query_as!(
+                TestAuthor,
+                "SELECT id, name, biography_id FROM authors WHERE id = ?",
+                created.id
+            )
+            .fetch_one(&pool)
+            .await
+            .unwrap();
+
+            assert_eq!(verified.id, created.id);
+            assert_eq!(verified.name, created.name);
+            assert_eq!(verified.biography_id, created.biography_id);
+        }
+    }
+}
diff --git a/tests/sqlite/fixtures/simple_struct.sql b/tests/sqlite/fixtures/simple_struct.sql
new file mode 100644
index 0000000..3da6929
--- /dev/null
+++ b/tests/sqlite/fixtures/simple_struct.sql
@@ -0,0 +1,9 @@
+INSERT INTO biographies (content)
+VALUES ('Some text'),
+       ('Some other text'),
+       ('Biography for no one');
+
+INSERT INTO authors (name, biography_id)
+VALUES ('J.R.R. Tolkien', 2),
+       ('George Orwell', NULL),
+       ('Jack London', 1);
diff --git a/tests/sqlite/models.rs b/tests/sqlite/models.rs
new file mode 100644
index 0000000..d229b61
--- /dev/null
+++ b/tests/sqlite/models.rs
@@ -0,0 +1,46 @@
+#![cfg(feature = "sqlite")]
+
+use georm::Georm;
+
+#[derive(Debug, Georm, PartialEq, Eq, Default)]
+#[georm(
+    table = "biographies",
+    one_to_one = [{
+        name = "author", remote_id = "biography_id", table = "authors", entity = Author
+    }]
+)]
+pub struct Biography {
+    #[georm(id)]
+    pub id: i64,
+    pub content: String,
+}
+
+#[derive(Debug, Georm, PartialEq, Eq, Default)]
+#[georm(table = "authors")]
+pub struct Author {
+    #[georm(id)]
+    pub id: i64,
+    pub name: String,
+    #[georm(relation = {entity = Biography, table = "biographies", name = "biography", nullable = true})]
+    pub biography_id: Option<i64>,
+}
+
+impl PartialOrd for Author {
+    fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
+        Some(self.id.cmp(&other.id))
+    }
+}
+
+impl Ord for Author {
+    fn cmp(&self, other: &Self) -> std::cmp::Ordering {
+        self.id.cmp(&other.id)
+    }
+}
+
+// UserRole (composite key, chrono::DateTime<Utc> column) is deliberately not
+// ported here yet: SQLite's compile-time query macros infer TEXT columns as
+// `String`, not `chrono::DateTime<Utc>` — unlike Postgres's TIMESTAMPTZ, which
+// maps directly. Fixing this needs per-column `query_as!` type overrides
+// (`"assigned_at as \"assigned_at: DateTime<Utc>\""`) threaded through the
+// dialect's generated queries, which is follow-up work, not part of this
+// first SQLite slice. See tests/sqlite/composite_key.rs (not yet added).
diff --git a/tests/sqlite/simple_struct.rs b/tests/sqlite/simple_struct.rs
new file mode 100644
index 0000000..fc28e3c
--- /dev/null
+++ b/tests/sqlite/simple_struct.rs
@@ -0,0 +1,159 @@
+#![cfg(feature = "sqlite")]
+
+use georm::Georm;
+use rand::seq::SliceRandom;
+
+use models::Author;
+mod models;
+
+#[sqlx::test(migrations = "./migrations/sqlite", fixtures("simple_struct"))]
+async fn find_all_query_works(pool: sqlx::SqlitePool) -> sqlx::Result<()> {
+    let result = Author::find_all(&pool).await?;
+    assert_eq!(3, result.len());
+    Ok(())
+}
+
+#[sqlx::test(migrations = "./migrations/sqlite")]
+async fn find_all_returns_empty_vec_on_empty_table(pool: sqlx::SqlitePool) -> sqlx::Result<()> {
+    let result = Author::find_all(&pool).await?;
+    assert_eq!(0, result.len());
+    Ok(())
+}
+
+#[sqlx::test(migrations = "./migrations/sqlite", fixtures("simple_struct"))]
+async fn find_query_works(pool: sqlx::SqlitePool) -> sqlx::Result<()> {
+    let id = 1;
+    let res = Author::find(&pool, &id).await?;
+    assert!(res.is_some());
+    let res = res.unwrap();
+    assert_eq!(String::from("J.R.R. Tolkien"), res.name);
+    assert_eq!(1, res.id);
+    Ok(())
+}
+
+#[sqlx::test(migrations = "./migrations/sqlite")]
+async fn find_returns_none_if_not_found(pool: sqlx::SqlitePool) -> sqlx::Result<()> {
+    let res = Author::find(&pool, &420).await?;
+    assert!(res.is_none());
+    Ok(())
+}
+
+#[sqlx::test(migrations = "./migrations/sqlite")]
+async fn create_works(pool: sqlx::SqlitePool) -> sqlx::Result<()> {
+    let author = Author {
+        id: 1,
+        name: "J.R.R. Tolkien".into(),
+        ..Default::default()
+    };
+    author.create(&pool).await?;
+    let all_authors = Author::find_all(&pool).await?;
+    assert_eq!(1, all_authors.len());
+    assert_eq!(vec![author], all_authors);
+    Ok(())
+}
+
+#[sqlx::test(migrations = "./migrations/sqlite", fixtures("simple_struct"))]
+async fn create_fails_if_already_exists(pool: sqlx::SqlitePool) -> sqlx::Result<()> {
+    let author = Author {
+        id: 2,
+        name: "Miura Kentaro".into(),
+        ..Default::default()
+    };
+    let result = author.create(&pool).await;
+    assert!(result.is_err());
+    Ok(())
+}
+
+#[sqlx::test(migrations = "./migrations/sqlite", fixtures("simple_struct"))]
+async fn update_works(pool: sqlx::SqlitePool) -> sqlx::Result<()> {
+    let expected_initial = Author {
+        name: "J.R.R. Tolkien".into(),
+        id: 1,
+        biography_id: Some(2),
+    };
+    let expected_final = Author {
+        name: "Jolkien Rolkien Rolkien Tolkien".into(),
+        id: 1,
+        biography_id: Some(2),
+    };
+    let tolkien = Author::find(&pool, &1).await?;
+    assert!(tolkien.is_some());
+    let mut tolkien = tolkien.unwrap();
+    assert_eq!(expected_initial, tolkien);
+    tolkien.name = expected_final.name.clone();
+    let updated = tolkien.update(&pool).await?;
+    assert_eq!(expected_final, updated);
+    Ok(())
+}
+
+#[sqlx::test(migrations = "./migrations/sqlite")]
+async fn update_fails_if_not_already_exists(pool: sqlx::SqlitePool) -> sqlx::Result<()> {
+    let author = Author {
+        id: 2,
+        name: "Miura Kentaro".into(),
+        ..Default::default()
+    };
+    let result = author.update(&pool).await;
+    assert!(result.is_err());
+    Ok(())
+}
+
+#[sqlx::test(migrations = "./migrations/sqlite")]
+async fn should_create_if_does_not_exist(pool: sqlx::SqlitePool) -> sqlx::Result<()> {
+    let all_authors = Author::find_all(&pool).await?;
+    assert_eq!(0, all_authors.len());
+    let author = Author {
+        id: 4,
+        name: "Miura Kentaro".into(),
+        ..Default::default()
+    };
+    author.upsert(&pool).await?;
+    let all_authors = Author::find_all(&pool).await?;
+    assert_eq!(1, all_authors.len());
+    Ok(())
+}
+
+#[sqlx::test(migrations = "./migrations/sqlite", fixtures("simple_struct"))]
+async fn should_update_if_exist(pool: sqlx::SqlitePool) -> sqlx::Result<()> {
+    let all_authors = Author::find_all(&pool).await?;
+    assert_eq!(3, all_authors.len());
+    let author = Author {
+        id: 2,
+        name: "Miura Kentaro".into(),
+        ..Default::default()
+    };
+    author.upsert(&pool).await?;
+    let mut all_authors = Author::find_all(&pool).await?;
+    all_authors.sort();
+    assert_eq!(3, all_authors.len());
+    assert_eq!(author, all_authors[1]);
+    Ok(())
+}
+
+#[sqlx::test(migrations = "./migrations/sqlite", fixtures("simple_struct"))]
+async fn delete_by_id_should_delete_only_one_entry(pool: sqlx::SqlitePool) -> sqlx::Result<()> {
+    let id = 2;
+    let all_authors = Author::find_all(&pool).await?;
+    assert_eq!(3, all_authors.len());
+    assert!(all_authors.iter().any(|author| author.get_id() == id));
+    let result = Author::delete_by_id(&pool, &id).await?;
+    assert_eq!(1, result);
+    let all_authors = Author::find_all(&pool).await?;
+    assert_eq!(2, all_authors.len());
+    assert!(all_authors.iter().all(|author| author.get_id() != id));
+    Ok(())
+}
+
+#[sqlx::test(migrations = "./migrations/sqlite", fixtures("simple_struct"))]
+async fn delete_should_delete_current_entity_from_db(pool: sqlx::SqlitePool) -> sqlx::Result<()> {
+    let mut all_authors = Author::find_all(&pool).await?;
+    assert_eq!(3, all_authors.len());
+    all_authors.shuffle(&mut rand::rng());
+    let author = all_authors.first().unwrap();
+    let result = author.delete(&pool).await?;
+    assert_eq!(1, result);
+    let all_authors = Author::find_all(&pool).await?;
+    assert_eq!(2, all_authors.len());
+    assert!(all_authors.iter().all(|a| a.get_id() != author.get_id()));
+    Ok(())
+}