diff --git a/app/models/recon_result.py b/app/models/recon_result.py
new file mode 100644
index 0000000..43c4a77
--- /dev/null
+++ b/app/models/recon_result.py
@@ -0,0 +1,347 @@
+import logging
+from typing import Optional, Annotated, Union, Type, Any
+from datetime import date, datetime
+import xml.etree.ElementTree as ET
+import math
+
+
+from pydantic import BaseModel, ConfigDict, Field, WithJsonSchema, field_serializer
+import pandas as pd
+import numpy as np
+
+from app.core.config import config
+from app.core.refdata import ReconResultStates
+from app.core.exceptions import ReconExecption
+from app.engine.loaders import url_to_schema
+from app.core.file_util import url_to_filepath
+
+DataFrame = Annotated[
+    pd.DataFrame,
+    WithJsonSchema({'type': 'object', 'additionalProperties': 'str'}),
+]
+
+
+NaTType: Type[Any] = type(pd.NaT)
+
+sanitized_model_config: dict[str, Any] = {
+    "arbitrary_types_allowed": True,
+    "json_encoders": {
+        NaTType: lambda v: None,
+        np.integer: int,
+        np.floating: lambda v: None if (math.isnan(v) or math.isinf(v)) else float(v),
+        np.bool_: bool,
+        np.ndarray: lambda v: v.tolist(),
+        pd.Timestamp: lambda v: v.isoformat(),
+        pd.Timedelta: lambda v: v.total_seconds(),
+        pd.Period: str,
+        float: lambda v: None if (math.isnan(v) or math.isinf(v)) else v,
+    }
+}
+
+
+def sanitize_for_json(obj: Any) -> Any:
+    """Recursively convert Pandas/NumPy types into JSON-safe Python types."""
+    if obj is None:
+        return None
+    if obj is pd.NaT or isinstance(obj, NaTType):
+        return None
+    if isinstance(obj, (np.integer,)):
+        return int(obj)
+    if isinstance(obj, (np.floating,)):
+        return None if (math.isnan(obj) or math.isinf(obj)) else float(obj)
+    if isinstance(obj, (np.bool_,)):
+        return bool(obj)
+    if isinstance(obj, (np.ndarray,)):
+        return [sanitize_for_json(x) for x in obj.tolist()]
+    if isinstance(obj, pd.Timestamp):
+        return obj.isoformat()
+    if isinstance(obj, pd.Timedelta):
+        return obj.total_seconds()
+    if isinstance(obj, pd.Period):
+        return str(obj)
+    if isinstance(obj, (list, tuple, set)):
+        return [sanitize_for_json(x) for x in obj]
+    if isinstance(obj, dict):
+        return {k: sanitize_for_json(v) for k, v in obj.items()}
+    return obj
+
+
+class Profile(BaseModel):
+    model_config = ConfigDict(**sanitized_model_config)
+
+    url: str
+    name: str
+    row_count: int
+    total_nulls: int
+    threshold_test_results: dict[str,str] = {}
+    query: Optional[str] = None
+    table_file: Optional[str] = None
+
+
+from typing import Optional
+from pydantic import BaseModel, Field
+
+
+class ReconResultValue(BaseModel):
+    column_name: str
+    column_value_raw: Any = Field(alias="_column_value")
+    ordinal_position: int
+    data_type: str
+    character_maximum_length: int
+    numeric_precision: int
+    numeric_scale: int
+    allow_null: int
+
+    def __init__(
+        self,
+        column_name: str,
+        column_value_raw: Any,
+        ordinal_position: int,
+        data_type: str,
+        character_maximum_length: int,
+        numeric_precision: int,
+        numeric_scale: int,
+        allow_null: int
+    ):
+        # Pass values to BaseModel using keyword args for validation
+        super().__init__(
+            column_name=column_name,
+            _column_value=column_value_raw,
+            ordinal_position=ordinal_position,
+            data_type=data_type,
+            character_maximum_length=character_maximum_length,
+            numeric_precision=numeric_precision,
+            numeric_scale=numeric_scale,
+            allow_null=allow_null
+        )
+
+    @property
+    def column_value(self) -> str:
+        """Return processed column value."""
+        val = self.column_value_raw
+
+        if val is None:
+            if not self.allow_null:
+                logging.warning(f"Result published with empty value for non-nullable field: {self.column_name}")
+                return "Undefined"
+            return ""
+
+        s = str(val)
+        s = s.replace('\t', ' ').replace('\n', ' ').replace('\r', ' ')
+        while '  ' in s:
+            s = s.replace('  ', ' ')
+        s = s.strip()
+
+        if self.character_maximum_length > 0 and len(s) > self.character_maximum_length:
+            return s[:self.character_maximum_length]
+
+        return s
+
+
+
+class ReconResult(BaseModel):
+    model_config = ConfigDict(**sanitized_model_config)
+    
+    recon_job_id: int
+    recon_job_name: str
+    recon_config_reference: str
+    recon_config_name: str
+    business_process: str
+    as_at_date: date
+    data_type: str
+    frequency: str
+    start_timestamp: datetime = Field(default_factory=datetime.now)
+    end_timestamp: Optional[datetime] = None
+    recon_state: Optional[ReconResultStates] = None
+    source_profiles: list[Profile] = []
+    dest_profile: Optional[Profile] = None
+
+    problems: dict[str,dict|str] = {}
+    @field_serializer("problems")
+    def serialize_problems(self, problems: dict, _info):
+        """Automatically sanitize problems before JSON serialization."""
+        return sanitize_for_json(problems)
+
+    environment: str = ""
+
+    pattern: str
+    pattern_result: Optional[bool] = None
+    pattern_failed_row_count: int = 0
+    pattern_rows_reconciled: int = 0
+
+    username: Optional[str] = None
+    due_timestamp: Optional[datetime] = None
+
+    _result_values: list[list[ReconResultValue]] = []
+    @property
+    def result_values(self):
+        if self._result_values == []:
+            self.create_recon_result_values()
+        return self._result_values
+
+
+    _pattern_result_description: Optional[str] = None
+    @property
+    def pattern_result_description(self) -> Optional[str]:
+        """
+        Escaped special characters like newlines, tabs, etc.
+        """
+        if self._pattern_result_description is None:
+            return None
+        return self._pattern_result_description.encode('unicode_escape').decode('utf-8')
+
+    @pattern_result_description.setter
+    def pattern_result_description(self, value: Optional[str]) -> None:
+        if value is not None and not isinstance(value, str):
+            raise TypeError("pattern_result_description must be a string or None")
+        self._pattern_result_description = value
+
+
+    def create_recon_result_values(self) -> None:
+        if self.dest_profile is None:
+            raise ReconExecption("No dest profile created")
+        if self.recon_state is None:
+            raise ReconExecption("Unexpeded null recon_state")
+
+        target_schema = url_to_schema(self.dest_profile.url)
+
+        for source in self.source_profiles:
+            row_values:list[ReconResultValue] = []
+            source_schema = url_to_schema(source.url)
+
+            row_values.append( ReconResultValue("Environment", f"{self.environment}", 1, "varchar", 100, 0, 0, 0))
+            row_values.append( ReconResultValue("Rec Number", f"{self.recon_job_id}", 2, "int", 0, 10, 0, 0))
+            row_values.append( ReconResultValue("Business Ref", f"{self.recon_config_name}", 3, "varchar", 200, 0, 0, 0))
+            row_values.append( ReconResultValue("Business Data Type", f"{self.data_type}", 4, "varchar", 200, 0, 0, 0))
+            row_values.append( ReconResultValue("Business Process", f"{self.business_process}", 5, "varchar", 200, 0, 0, 0))
+            row_values.append( ReconResultValue("Config Reference", f"{self.recon_config_reference}", 6, "varchar", 200, 0, 0, 0))
+            row_values.append( ReconResultValue("Recon Date", f"{self.as_at_date.isoformat()}", 7, "datetime", 0, 0, 0, 0))
+            row_values.append( ReconResultValue("Run Date", f"{self.start_timestamp.isoformat()}", 8, "datetime", 0, 0, 0, 0))
+            row_values.append( ReconResultValue("Frequency", f"{self.frequency}", 9, "varchar", 30, 0, 0, 0))
+            row_values.append( ReconResultValue("Overall Status", f"{self.recon_state.value}", 10, "varchar", 150, 0, 0, 0))
+            row_values.append( ReconResultValue("Data Reconciled", f"{self.pattern_result}".lower(), 11, "boolean", 0, 0, 0, 0))
+            row_values.append( ReconResultValue("Data Recon Reason", f"{self.pattern_result_description}", 12, "varchar", 1024, 0, 0, 0))
+            row_values.append( ReconResultValue("Records Failed Count", f"{self.pattern_failed_row_count}", 13, "int", 0, 10, 0, 0))
+            row_values.append( ReconResultValue("Source Profile Results", f"{source.threshold_test_results}", 14, "varchar", 512, 0, 0, 0))
+            row_values.append( ReconResultValue("Source System/DB", f"{source.name}", 15, "varchar", 512, 0, 0, 0))
+            row_values.append( ReconResultValue("Source Schema", f"{source_schema}", 16, "varchar", 512, 0, 0, 1))
+            row_values.append( ReconResultValue("Source Table/File", f"{source.table_file}", 17, "varchar", 512, 0, 0, 0))
+            row_values.append( ReconResultValue("Source URL", f"{source.url}", 18, "varchar", 512, 0, 0, 0))
+            row_values.append( ReconResultValue("Source Records", f"{source.row_count}", 19, "int", 0, 0, 0, 0))
+            row_values.append( ReconResultValue("Target Profile Results", f"{self.dest_profile.threshold_test_results}", 20, "varchar", 512, 0, 0, 0))
+            row_values.append( ReconResultValue("Target System/DB", f"{self.dest_profile.name}", 21, "varchar", 512, 0, 0, 0))
+            row_values.append( ReconResultValue("Target Schema", f"{target_schema}", 22, "varchar", 512, 0, 0, 1))
+            row_values.append( ReconResultValue("Target Table/File", f"{self.dest_profile.table_file}", 23, "varchar", 512, 0, 0, 0))
+            row_values.append( ReconResultValue("Target URL", f"{self.dest_profile.url}", 24, "varchar", 512, 0, 0, 0))
+            row_values.append( ReconResultValue("Target Records", f"{self.dest_profile.row_count}", 25, "int", 0, 10, 0, 0))
+            row_values.append( ReconResultValue("Rows Reconciled", f"{self.pattern_rows_reconciled}", 26, "int", 0, 10, 0, 0))
+            row_values.append( ReconResultValue("Start Timestamp", f"{self.start_timestamp.isoformat()}", 27, "datetime", 0, 0, 0, 1))
+            row_values.append( ReconResultValue("End Timestamp", self.end_timestamp.isoformat() if self.end_timestamp else "", 28, "datetime", 0, 0, 0, 1))
+
+            row_values.append( ReconResultValue("Username", self.username if self.username else "", 29, "varchar", 256, 0, 0, 1))
+            row_values.append( ReconResultValue("Due Timestamp", self.due_timestamp.isoformat() if self.due_timestamp else "", 30, "datetime", 0, 0, 0, 1))
+            row_values.append( ReconResultValue("Pattern", self.pattern if self.pattern else "", 31, "varchar", 128, 0, 0, 1))
+            row_values.append( ReconResultValue("Source Filter", source.query if source.query else "", 32, "varchar", 1024, 0, 0, 1))
+            row_values.append( ReconResultValue("Target Filter", self.dest_profile.query if self.dest_profile.query else "", 32, "varchar", 1024, 0, 0, 1))
+
+            self._result_values.append(row_values)
+
+    
+    def to_csv(self, result_url):
+        if result_url.startswith("file:") != True:
+            raise ReconExecption(f"Only file system urls are allowed for the result path so far. Config has {result_url}")
+        
+        # Create the csv dictionaries to be passed to writer
+        csv_results:list[dict] = []
+        for row in self.result_values:
+            csv_row = {}
+            for column in row:
+                csv_row[column.column_name] = column.column_value
+            csv_results.append(csv_row)
+
+        # generate filenames
+        output_file_name = f"reconciliation_results_{datetime.now().strftime('%Y%m%d_%H%M%S')}.dat"
+        result_filename = url_to_filepath(result_url + output_file_name)
+        control_file_name = f"{result_filename}.xml"
+
+        # Write csv data file
+        df = pd.DataFrame(csv_results)
+        delimiter = "|~"
+
+        with open(result_filename, "w", encoding="utf-8") as f:
+            # Write header
+            f.write(delimiter.join(df.columns) + "\n")
+            # Write rows
+            for row in df.itertuples(index=False):
+                f.write(delimiter.join(map(str, row)) + "\n")
+        logging.info(f"Wrote results file {result_filename} with {len(df)} rows")
+
+        # Create the csv control file data
+        control_file_content = self.output_control_file(
+                                    file_name=output_file_name,
+                                    number_of_records=len(csv_results),
+                                    legacy_feed_date=self.start_timestamp.date(),
+                                    delimiter=delimiter
+                                )
+        # Write csv control file
+        with open(control_file_name, mode='w', newline='') as file:
+            file.write(control_file_content)
+        logging.info(f"Wrote results control file {control_file_name}")
+
+
+    def output_control_file(self, file_name: str, number_of_records: int, legacy_feed_date: Union[str, date], delimiter) -> str:
+        
+        if isinstance(legacy_feed_date, date):
+            legacy_feed_date = legacy_feed_date.isoformat()
+        else:
+            legacy_feed_date = str(legacy_feed_date)
+
+        root = ET.Element("Feed")
+
+        def add(tag: str, text: str) -> None:
+            el = ET.SubElement(root, tag)
+            el.text = text
+
+        # Header (hardcoded to match example)
+        add("File", file_name)
+        add("ExtractProg", "RECONRANGER")
+        add("GenerationDate", f"{datetime.now().strftime('%b %d %Y %I:%M%p').upper()}")
+        add("NumberOfRecords", str(number_of_records))
+        add("LegacyFeedDate", legacy_feed_date)
+        add("FieldDelimiter", delimiter)
+        add("ContactEmailAddress", config.contact_email)
+
+        data_schema = ET.SubElement(root, "DataSchema")
+        cols = ET.SubElement(data_schema, "COLS")
+
+        def add_col(
+            column_name: str,
+            ordinal_position: int,
+            data_type: str,
+            character_maximum_length: int,
+            numeric_precision: int,
+            numeric_scale: int,
+            allow_null: int,
+        ) -> None:
+            col = ET.SubElement(cols, "COL")
+            ET.SubElement(col, "COLUMN_NAME").text = column_name
+            ET.SubElement(col, "ORDINAL_POSITION").text = str(ordinal_position)
+            ET.SubElement(col, "DATA_TYPE").text = data_type
+            ET.SubElement(col, "CHARACTER_MAXIMUM_LENGTH").text = str(character_maximum_length)
+            ET.SubElement(col, "NUMERIC_PRECISION").text = str(numeric_precision)
+            ET.SubElement(col, "NUMERIC_SCALE").text = str(numeric_scale)
+            ET.SubElement(col, "ALLOW_NULL").text = str(allow_null)
+
+        for column in self.result_values[0]:
+            add_col(
+                    column.column_name.replace(" ","_").replace("/","_"),
+                    column.ordinal_position, 
+                    column.data_type, 
+                    column.character_maximum_length, 
+                    column.numeric_precision, 
+                    column.numeric_scale, 
+                    column.allow_null
+            )
+
+        xml_bytes = ET.tostring(root, encoding="utf-8", xml_declaration=False)
+        return xml_bytes.decode("utf-8")            
\ No newline at end of file