diff --git a/regex_humanizer/parser/ast.py b/regex_humanizer/parser/ast.py
new file mode 100644
index 0000000..81f8c82
--- /dev/null
+++ b/regex_humanizer/parser/ast.py
@@ -0,0 +1,152 @@
+"""Abstract Syntax Tree nodes for regex patterns."""
+
+from abc import ABC, abstractmethod
+from dataclasses import dataclass, field
+from typing import List, Optional, Union
+
+
+@dataclass
+class ASTNode(ABC):
+    """Base class for all AST nodes."""
+
+    position: int = 0
+
+    @abstractmethod
+    def to_regex(self) -> str:
+        """Convert the node back to a regex pattern."""
+        pass
+
+
+@dataclass
+class Literal(ASTNode):
+    """A literal character or string."""
+
+    escaped: bool = False
+    value: str = ""
+
+    def to_regex(self) -> str:
+        if self.escaped or not self.value.isalnum():
+            escaped = ""
+            for char in self.value:
+                if char in r".^$*+?{}[]\|()":
+                    escaped += "\\" + char
+                else:
+                    escaped += char
+            return escaped
+        return self.value
+
+
+@dataclass
+class CharacterClass(ASTNode):
+    """A character class like [a-z] or [^A-Z]."""
+
+    inverted: bool = False
+    characters: List[str] = field(default_factory=list)
+    ranges: List[tuple] = field(default_factory=list)
+
+    def to_regex(self) -> str:
+        content = ""
+        special_chars = {'^', '-', ']', '\\'}
+        for char in self.characters:
+            if char in special_chars:
+                content += "\\" + char
+            else:
+                content += char
+        for start, end in self.ranges:
+            content += f"{start}-{end}"
+        negated = "^" if self.inverted else ""
+        return f"[{negated}{content}]"
+
+
+@dataclass
+class Quantifier(ASTNode):
+    """A quantifier like *, +, ?, {n}, {n,m}."""
+
+    MIN_UNBOUNDED = -1
+    MAX_UNBOUNDED = -1
+
+    min: int = 0
+    max: int = 1
+    lazy: bool = False
+    possessive: bool = False
+    child: Optional[ASTNode] = None
+
+    def to_regex(self) -> str:
+        base = ""
+        if self.min == 0 and self.max == 1:
+            base = "?"
+        elif self.min == 0 and self.max == self.MAX_UNBOUNDED:
+            base = "*"
+        elif self.min == 1 and self.max == self.MAX_UNBOUNDED:
+            base = "+"
+        elif self.min == self.max:
+            base = f"{{{self.min}}}"
+        elif self.max == self.MAX_UNBOUNDED:
+            base = f"{{{self.min},}}"
+        else:
+            base = f"{{{self.min},{self.max}}}"
+        modifier = "?" if self.lazy else ""
+        modifier = "+" if self.possessive else modifier
+        return base + modifier
+
+
+@dataclass
+class Group(ASTNode):
+    """A capturing or non-capturing group."""
+
+    content: List[ASTNode] = field(default_factory=list)
+    capturing: bool = True
+    name: Optional[str] = None
+
+    def to_regex(self) -> str:
+        inner = "".join(node.to_regex() for node in self.content)
+        if self.name:
+            return f"(?P<{self.name}>{inner})"
+        elif not self.capturing:
+            return f"(?:{inner})"
+        return f"({inner})"
+
+
+@dataclass
+class Alternation(ASTNode):
+    """An alternation (OR) construct."""
+
+    options: List[List[ASTNode]] = field(default_factory=list)
+
+    def to_regex(self) -> str:
+        option_strs = []
+        for option in self.options:
+            option_strs.append("".join(node.to_regex() for node in option))
+        return "|".join(option_strs)
+
+
+@dataclass
+class Anchor(ASTNode):
+    """An anchor like ^, $, \\b, \\B."""
+
+    kind: str = "^"
+
+    def to_regex(self) -> str:
+        return self.kind
+
+
+@dataclass
+class SpecialSequence(ASTNode):
+    """A special sequence like \\d, \\w, \\s, etc."""
+
+    sequence: str = "\\d"
+
+    def to_regex(self) -> str:
+        return self.sequence
+
+
+@dataclass
+class Backreference(ASTNode):
+    """A backreference like \\1 or \\k<name>."""
+
+    reference: Union[int, str] = 1
+
+    def to_regex(self) -> str:
+        if isinstance(self.reference, str):
+            return f"\\k<{self.reference}>"
+        return f"\\{self.reference}"