diff --git a/nac3core/src/codegen/expr.rs b/nac3core/src/codegen/expr.rs
index 3ca72835..98972a81 100644
--- a/nac3core/src/codegen/expr.rs
+++ b/nac3core/src/codegen/expr.rs
@@ -10,7 +10,10 @@ use crate::{
     },
     symbol_resolver::{SymbolValue, ValueEnum},
     toplevel::{DefinitionId, TopLevelDef},
-    typecheck::typedef::{FunSignature, FuncArg, Type, TypeEnum, Unifier},
+    typecheck::{
+        typedef::{FunSignature, FuncArg, Type, TypeEnum, Unifier},
+        magic_methods::{binop_name, binop_assign_name},
+    },
 };
 use inkwell::{
     AddressSpace,
@@ -927,21 +930,29 @@ pub fn gen_binop_expr<'ctx, 'a, G: CodeGenerator>(
     left: &Expr<Option<Type>>,
     op: &Operator,
     right: &Expr<Option<Type>>,
-) -> Result<ValueEnum<'ctx>, String> {
+    loc: Location,
+    is_aug_assign: bool,
+) -> Result<Option<ValueEnum<'ctx>>, String> {
     let ty1 = ctx.unifier.get_representative(left.custom.unwrap());
     let ty2 = ctx.unifier.get_representative(right.custom.unwrap());
-    let left = generator.gen_expr(ctx, left)?.unwrap().to_basic_value_enum(ctx, generator, left.custom.unwrap())?;
-    let right = generator.gen_expr(ctx, right)?.unwrap().to_basic_value_enum(ctx, generator, right.custom.unwrap())?;
+    let left_val = generator
+        .gen_expr(ctx, left)?
+        .unwrap()
+        .to_basic_value_enum(ctx, generator, left.custom.unwrap())?;
+    let right_val = generator
+        .gen_expr(ctx, right)?
+        .unwrap()
+        .to_basic_value_enum(ctx, generator, right.custom.unwrap())?;
 
     // we can directly compare the types, because we've got their representatives
     // which would be unchanged until further unification, which we would never do
     // when doing code generation for function instances
-    Ok(if ty1 == ty2 && [ctx.primitives.int32, ctx.primitives.int64].contains(&ty1) {
-        ctx.gen_int_ops(generator, op, left, right, true)
+    if ty1 == ty2 && [ctx.primitives.int32, ctx.primitives.int64].contains(&ty1) {
+        Ok(Some(ctx.gen_int_ops(generator, op, left_val, right_val, true).into()))
     } else if ty1 == ty2 && [ctx.primitives.uint32, ctx.primitives.uint64].contains(&ty1) {
-        ctx.gen_int_ops(generator, op, left, right, false)
+        Ok(Some(ctx.gen_int_ops(generator, op, left_val, right_val, false).into()))
     } else if ty1 == ty2 && ctx.primitives.float == ty1 {
-        ctx.gen_float_ops(op, left, right)
+        Ok(Some(ctx.gen_float_ops(op, left_val, right_val).into()))
     } else if ty1 == ctx.primitives.float && ty2 == ctx.primitives.int32 {
         // Pow is the only operator that would pass typecheck between float and int
         assert!(*op == Operator::Pow);
@@ -951,14 +962,68 @@ pub fn gen_binop_expr<'ctx, 'a, G: CodeGenerator>(
             let ty = f64_t.fn_type(&[f64_t.into(), i32_t.into()], false);
             ctx.module.add_function("llvm.powi.f64.i32", ty, None)
         });
-        ctx.builder
-            .build_call(pow_intr, &[left.into(), right.into()], "f_pow_i")
+        let res = ctx.builder
+            .build_call(pow_intr, &[left_val.into(), right_val.into()], "f_pow_i")
             .try_as_basic_value()
-            .unwrap_left()
+            .unwrap_left();
+        Ok(Some(res.into()))
     } else {
-        unimplemented!()
+        let (op_name, id) = if let TypeEnum::TObj { fields, obj_id, .. } =
+            ctx.unifier.get_ty_immutable(left.custom.unwrap()).as_ref()
+        {
+            let (binop_name, binop_assign_name) = (
+                binop_name(op).into(),
+                binop_assign_name(op).into()
+            );
+            // if is aug_assign, try aug_assign operator first
+            if is_aug_assign && fields.contains_key(&binop_assign_name) {
+                (binop_assign_name, *obj_id)
+            } else {
+                (binop_name, *obj_id)
+            }
+        } else {
+            unreachable!("must be tobj")
+        };
+        let signature = match ctx.calls.get(&loc.into()) {
+            Some(call) => ctx.unifier.get_call_signature(*call).unwrap(),
+            None => {
+                if let TypeEnum::TObj { fields, .. } =
+                    ctx.unifier.get_ty_immutable(left.custom.unwrap()).as_ref()
+                {
+                    let fn_ty = fields.get(&op_name).unwrap().0;
+                    if let TypeEnum::TFunc(sig) = ctx.unifier.get_ty_immutable(fn_ty).as_ref() {
+                        sig.clone()
+                    } else {
+                        unreachable!("must be func sig")
+                    }
+                } else {
+                    unreachable!("must be tobj")
+                }
+            },
+        };
+        let fun_id = {
+            let defs = ctx.top_level.definitions.read();
+            let obj_def = defs.get(id.0).unwrap().read();
+            if let TopLevelDef::Class { methods, .. } = &*obj_def {
+                let mut fun_id = None;
+                for (name, _, id) in methods.iter() {
+                    if name == &op_name {
+                        fun_id = Some(*id);
+                    }
+                }
+                fun_id.unwrap()
+            } else {
+                unreachable!()
+            }
+        };
+        generator
+            .gen_call(
+                ctx,
+                Some((left.custom.unwrap(), left_val.into())),
+                (&signature, fun_id),
+                vec![(None, right_val.into())],
+            ).map(|f| f.map(|f| f.into()))
     }
-    .into())
 }
 
 pub fn gen_expr<'ctx, 'a, G: CodeGenerator>(
@@ -1125,7 +1190,9 @@ pub fn gen_expr<'ctx, 'a, G: CodeGenerator>(
             phi.add_incoming(&[(&a, a_bb), (&b, b_bb)]);
             phi.as_basic_value().into()
         }
-        ExprKind::BinOp { op, left, right } => gen_binop_expr(generator, ctx, left, op, right)?,
+        ExprKind::BinOp { op, left, right } => {
+            return gen_binop_expr(generator, ctx, left, op, right, expr.location, false);
+        }
         ExprKind::UnaryOp { op, operand } => {
             let ty = ctx.unifier.get_representative(operand.custom.unwrap());
             let val =
diff --git a/nac3core/src/codegen/stmt.rs b/nac3core/src/codegen/stmt.rs
index 9209440d..5c1a7842 100644
--- a/nac3core/src/codegen/stmt.rs
+++ b/nac3core/src/codegen/stmt.rs
@@ -1020,8 +1020,8 @@ pub fn gen_stmt<'ctx, 'a, G: CodeGenerator>(
         StmtKind::For { .. } => generator.gen_for(ctx, stmt)?,
         StmtKind::With { .. } => generator.gen_with(ctx, stmt)?,
         StmtKind::AugAssign { target, op, value, .. } => {
-            let value = gen_binop_expr(generator, ctx, target, op, value)?;
-            generator.gen_assign(ctx, target, value)?;
+            let value = gen_binop_expr(generator, ctx, target, op, value, stmt.location, true)?;
+            generator.gen_assign(ctx, target, value.unwrap())?;
         }
         StmtKind::Try { .. } => gen_try(generator, ctx, stmt)?,
         StmtKind::Raise { exc, .. } => {
diff --git a/nac3core/src/typecheck/magic_methods.rs b/nac3core/src/typecheck/magic_methods.rs
index 36d46a6d..0a023b19 100644
--- a/nac3core/src/typecheck/magic_methods.rs
+++ b/nac3core/src/typecheck/magic_methods.rs
@@ -83,7 +83,7 @@ where
 
 pub fn impl_binop(
     unifier: &mut Unifier,
-    store: &PrimitiveStore,
+    _store: &PrimitiveStore,
     ty: Type,
     other_ty: &[Type],
     ret_ty: Type,
@@ -120,7 +120,7 @@ pub fn impl_binop(
             fields.insert(binop_assign_name(op).into(), {
                 (
                     unifier.add_ty(TypeEnum::TFunc(FunSignature {
-                        ret: store.none,
+                        ret: ret_ty,
                         vars: function_vars.clone(),
                         args: vec![FuncArg {
                             ty: other_ty,
diff --git a/nac3core/src/typecheck/type_inferencer/mod.rs b/nac3core/src/typecheck/type_inferencer/mod.rs
index 510a0f09..c72e1998 100644
--- a/nac3core/src/typecheck/type_inferencer/mod.rs
+++ b/nac3core/src/typecheck/type_inferencer/mod.rs
@@ -423,7 +423,7 @@ impl<'a> fold::Fold<()> for Inferencer<'a> {
                 (None, None) => {}
             },
             ast::StmtKind::AugAssign { target, op, value, .. } => {
-                let res_ty = self.infer_bin_ops(stmt.location, target, op, value)?;
+                let res_ty = self.infer_bin_ops(stmt.location, target, op, value, true)?;
                 self.unify(res_ty, target.custom.unwrap(), &stmt.location)?;
             }
             ast::StmtKind::Assert { test, msg, .. } => {
@@ -505,7 +505,7 @@ impl<'a> fold::Fold<()> for Inferencer<'a> {
             }
             ast::ExprKind::BoolOp { values, .. } => Some(self.infer_bool_ops(values)?),
             ast::ExprKind::BinOp { left, op, right } => {
-                Some(self.infer_bin_ops(expr.location, left, op, right)?)
+                Some(self.infer_bin_ops(expr.location, left, op, right, false)?)
             }
             ast::ExprKind::UnaryOp { op, operand } => Some(self.infer_unary_ops(op, operand)?),
             ast::ExprKind::Compare { left, ops, comparators } => {
@@ -1028,8 +1028,24 @@ impl<'a> Inferencer<'a> {
         left: &ast::Expr<Option<Type>>,
         op: &ast::Operator,
         right: &ast::Expr<Option<Type>>,
+        is_aug_assign: bool,
     ) -> InferenceResult {
-        let method = binop_name(op).into();
+        let method = if let TypeEnum::TObj { fields, .. } =
+            self.unifier.get_ty_immutable(left.custom.unwrap()).as_ref()
+        {
+            let (binop_name, binop_assign_name) = (
+                binop_name(op).into(),
+                binop_assign_name(op).into()
+            );
+            // if is aug_assign, try aug_assign operator first
+            if is_aug_assign && fields.contains_key(&binop_assign_name) {
+                binop_assign_name
+            } else {
+                binop_name
+            }
+        } else {
+            binop_name(op).into()
+        };
         self.build_method_call(
             location,
             method,
diff --git a/nac3standalone/demo/src/operators.py b/nac3standalone/demo/src/operators.py
new file mode 100644
index 00000000..0470b969
--- /dev/null
+++ b/nac3standalone/demo/src/operators.py
@@ -0,0 +1,257 @@
+from __future__ import annotations
+
+@extern
+def output_int32(x: int32):
+    ...
+@extern
+def output_uint32(x: uint32):
+    ...
+@extern
+def output_int64(x: int64):
+    ...
+@extern
+def output_uint64(x: uint64):
+    ...
+@extern
+def output_float64(x: float):
+    ...
+
+def run() -> int32:
+    test_int32()
+    test_uint32()
+    test_int64()
+    test_uint64()
+    test_A()
+    test_B()
+    return 0
+
+def test_int32():
+    a = 17
+    b = 3
+    output_int32(a + b)
+    output_int32(a - b)
+    output_int32(a * b)
+    output_int32(a // b)
+    output_int32(a % b)
+    output_int32(a | b)
+    output_int32(a ^ b)
+    output_int32(a & b)
+    output_int32(a << b)
+    output_int32(a >> b)
+    output_float64(a / b)
+    a += b
+    output_int32(a)
+    a -= b
+    output_int32(a)
+    a *= b
+    output_int32(a)
+    a //= b
+    output_int32(a)
+    a %= b
+    output_int32(a)
+    a |= b
+    output_int32(a)
+    a ^= b
+    output_int32(a)
+    a &= b
+    output_int32(a)
+    a <<= b
+    output_int32(a)
+    a >>= b
+    output_int32(a)
+    # fail because (a / b) is float
+    # a /= b
+
+def test_uint32():
+    a = uint32(17)
+    b = uint32(3)
+    output_uint32(a + b)
+    output_uint32(a - b)
+    output_uint32(a * b)
+    output_uint32(a // b)
+    output_uint32(a % b)
+    output_uint32(a | b)
+    output_uint32(a ^ b)
+    output_uint32(a & b)
+    output_uint32(a << b)
+    output_uint32(a >> b)
+    output_float64(a / b)
+    a += b
+    output_uint32(a)
+    a -= b
+    output_uint32(a)
+    a *= b
+    output_uint32(a)
+    a //= b
+    output_uint32(a)
+    a %= b
+    output_uint32(a)
+    a |= b
+    output_uint32(a)
+    a ^= b
+    output_uint32(a)
+    a &= b
+    output_uint32(a)
+    a <<= b
+    output_uint32(a)
+    a >>= b
+    output_uint32(a)
+
+def test_int64():
+    a = int64(17)
+    b = int64(3)
+    output_int64(a + b)
+    output_int64(a - b)
+    output_int64(a * b)
+    output_int64(a // b)
+    output_int64(a % b)
+    output_int64(a | b)
+    output_int64(a ^ b)
+    output_int64(a & b)
+    output_int64(a << b)
+    output_int64(a >> b)
+    output_float64(a / b)
+    a += b
+    output_int64(a)
+    a -= b
+    output_int64(a)
+    a *= b
+    output_int64(a)
+    a //= b
+    output_int64(a)
+    a %= b
+    output_int64(a)
+    a |= b
+    output_int64(a)
+    a ^= b
+    output_int64(a)
+    a &= b
+    output_int64(a)
+    a <<= b
+    output_int64(a)
+    a >>= b
+    output_int64(a)
+
+def test_uint64():
+    a = uint64(17)
+    b = uint64(3)
+    output_uint64(a + b)
+    output_uint64(a - b)
+    output_uint64(a * b)
+    output_uint64(a // b)
+    output_uint64(a % b)
+    output_uint64(a | b)
+    output_uint64(a ^ b)
+    output_uint64(a & b)
+    output_uint64(a << b)
+    output_uint64(a >> b)
+    output_float64(a / b)
+    a += b
+    output_uint64(a)
+    a -= b
+    output_uint64(a)
+    a *= b
+    output_uint64(a)
+    a //= b
+    output_uint64(a)
+    a %= b
+    output_uint64(a)
+    a |= b
+    output_uint64(a)
+    a ^= b
+    output_uint64(a)
+    a &= b
+    output_uint64(a)
+    a <<= b
+    output_uint64(a)
+    a >>= b
+    output_uint64(a)
+
+class A:
+    a: int32
+    def __init__(self, a: int32):
+        self.a = a
+    
+    def __add__(self, other: A) -> A:
+        output_int32(self.a + other.a)
+        return A(self.a + other.a)
+
+    def __sub__(self, other: A) -> A:
+        output_int32(self.a - other.a)
+        return A(self.a - other.a)
+
+def test_A():
+    a = A(17)
+    b = A(3)
+    
+    c = a + b
+    # fail due to alloca in __add__ function
+    # output_int32(c.a)
+
+    a += b
+    # fail due to alloca in __add__ function
+    # output_int32(a.a)
+    
+    a = A(17)
+    b = A(3)
+    d = a - b
+    # fail due to alloca in __add__ function
+    # output_int32(c.a)
+
+    a -= b
+    # fail due to alloca in __add__ function
+    # output_int32(a.a)
+
+    a = A(17)
+    b = A(3)
+    a.__add__(b)
+    a.__sub__(b)
+
+
+class B:
+    a: int32
+    def __init__(self, a: int32):
+        self.a = a
+    
+    def __add__(self, other: B) -> B:
+        output_int32(self.a + other.a)
+        return B(self.a + other.a)
+
+    def __sub__(self, other: B) -> B:
+        output_int32(self.a - other.a)
+        return B(self.a - other.a)
+    
+    def __iadd__(self, other: B) -> B:
+        output_int32(self.a + other.a + 24)
+        return B(self.a + other.a + 24)
+
+    def __isub__(self, other: B) -> B:
+        output_int32(self.a - other.a - 24)
+        return B(self.a - other.a - 24)
+
+def test_B():
+    a = B(17)
+    b = B(3)
+    
+    c = a + b
+    # fail due to alloca in __add__ function
+    # output_int32(c.a)
+
+    a += b
+    # fail due to alloca in __add__ function
+    # output_int32(a.a)
+    
+    a = B(17)
+    b = B(3)
+    d = a - b
+    # fail due to alloca in __add__ function
+    # output_int32(c.a)
+
+    a -= b
+    # fail due to alloca in __add__ function
+    # output_int32(a.a)
+
+    a = B(17)
+    b = B(3)
+    a.__add__(b)
+    a.__sub__(b)
\ No newline at end of file
diff --git a/nac3standalone/src/main.rs b/nac3standalone/src/main.rs
index 37925157..6b624aa6 100644
--- a/nac3standalone/src/main.rs
+++ b/nac3standalone/src/main.rs
@@ -205,6 +205,14 @@ fn main() {
             continue;
         }
 
+        // still needs to skip this `from __future__ import annotations` because this seems to be
+        // magic in python and there seems no way to patch it from another module..
+        if matches!(
+            &stmt.node,
+            StmtKind::ImportFrom { module, names, .. }
+                if module == &Some("__future__".into()) && names[0].name == "annotations".into()
+        ) { continue; }
+
         let (name, def_id, ty) =
             composer.register_top_level(stmt, Some(resolver.clone()), "__main__".into()).unwrap();