diff --git a/apps/api/src/dkv/dkv-parser.service.ts b/apps/api/src/dkv/dkv-parser.service.ts
new file mode 100644
index 0000000..f9237c8
--- /dev/null
+++ b/apps/api/src/dkv/dkv-parser.service.ts
@@ -0,0 +1,224 @@
+import { Injectable, Logger } from '@nestjs/common';
+import { PDFParse } from 'pdf-parse';
+import type { DkvVehicleBlock, DkvTransaction } from './dkv.types.js';
+
+/**
+ * DkvParserService
+ *
+ * Extracts structured vehicle/transaction data from DKV E-Rechnung PDF buffers.
+ *
+ * Security notes:
+ * - Parser operates only on buffers fetched by the inbox provider (T-07-01)
+ * - Calls destroy() to free PDF parser memory after extraction (T-07-01)
+ * - Generic error messages only on parse failure — no PDF content in logs (T-07-02)
+ *
+ * Wave 0 validation: the parsing logic here was empirically derived by running
+ * dkv-parser.validate.ts against user-files/invoice.pdf (April 2026, 27 vehicles).
+ * Result: 27 vehicle blocks, 66 transactions total.
+ *
+ * Two PDF extraction formats are handled:
+ * 1. Single-transaction (tab-separated): one row per transaction in the PDF
+ * 2. Multi-transaction (columnar): all values per column stacked vertically in the extracted text
+ */
+@Injectable()
+export class DkvParserService {
+  private readonly logger = new Logger(DkvParserService.name);
+
+  /**
+   * Parse a DKV E-Rechnung PDF buffer into structured vehicle blocks.
+   *
+   * @param buffer - Raw PDF bytes (from email attachment)
+   * @returns Array of vehicle blocks with transactions
+   * @throws Error with generic message if no vehicle blocks are found
+   */
+  async parsePdf(buffer: Buffer): Promise<DkvVehicleBlock[]> {
+    let text: string;
+    try {
+      text = await this.extractText(buffer);
+    } catch (err) {
+      this.logger.error('DKV PDF text extraction failed');
+      throw new Error('DKV invoice parsing failed — text extraction error');
+    }
+
+    const vehicles = this.parseDkvText(text);
+
+    if (vehicles.length === 0) {
+      this.logger.warn('DKV PDF parsed but yielded zero vehicle blocks');
+      throw new Error('DKV invoice parsing failed — no vehicle data found');
+    }
+
+    const totalTx = vehicles.reduce((s, v) => s + v.transactions.length, 0);
+    this.logger.log(
+      `DKV PDF parsed: ${vehicles.length} vehicle(s), ${totalTx} transaction(s)`,
+    );
+
+    return vehicles;
+  }
+
+  // ─── Private: PDF text extraction ──────────────────────────────────────────
+
+  private async extractText(buffer: Buffer): Promise<string> {
+    // pdf-parse v2 class-based API — do NOT use v1 pdfParse(buffer) function call
+    const parser = new PDFParse({ data: buffer });
+    const result = await parser.getText();
+    await parser.destroy(); // always free memory (T-07-01)
+    return result.text;
+  }
+
+  // ─── Private: Vehicle block parser ─────────────────────────────────────────
+
+  private parseDkvText(text: string): DkvVehicleBlock[] {
+    const vehicles: DkvVehicleBlock[] = [];
+
+    // Anchor on VEHICLE: marker — each block extends until next VEHICLE: or end
+    // Kennzeichen format: "GP-JL 728E", "GP ML 720", etc.
+    const vehicleBlockPattern =
+      /VEHICLE:\s+([A-Z0-9 ._\-]+?)\s+CARD NO\.:\s+(\S+)([\s\S]*?)(?=VEHICLE:|$)/g;
+
+    let match: RegExpExecArray | null;
+    while ((match = vehicleBlockPattern.exec(text)) !== null) {
+      const kennzeichen = match[1].trim();
+      const cardNumber = match[2].trim();
+      const blockText = match[3];
+      const transactions = this.parseTransactionRows(blockText);
+      vehicles.push({ kennzeichen, cardNumber, transactions });
+    }
+
+    return vehicles;
+  }
+
+  // ─── Private: Transaction row dispatcher ───────────────────────────────────
+
+  private parseTransactionRows(blockText: string): DkvTransaction[] {
+    const lines = blockText
+      .split('\n')
+      .map(l => l.trim())
+      .filter(l => l && !l.startsWith('»'));
+
+    if (lines.length === 0) return [];
+
+    const datePattern = /^\d{2}\.\d{2}\.\d{4}/;
+
+    // If any date-starting line also contains a tab → single-tx tab format
+    const hasTabbedDateLine = lines.some(l => datePattern.test(l) && l.includes('\t'));
+
+    if (hasTabbedDateLine) {
+      return lines
+        .filter(l => datePattern.test(l) && l.includes('\t'))
+        .map(line => this.parseSingleTxLine(line))
+        .filter((tx): tx is DkvTransaction => tx !== null);
+    }
+
+    // Multi-transaction columnar format
+    return this.parseMultiTxColumnar(lines);
+  }
+
+  // ─── Private: Single-transaction tab-separated line ────────────────────────
+  //
+  // DKV single-tx tab format (empirically verified against invoice.pdf):
+  //
+  //   [0]  Lieferdatum      DD.MM.YYYY
+  //   [1]  Station Name     e.g. "ESSO"
+  //   [2]  Ort              e.g. "DONZDORF"
+  //   [3]  Servicest.-Nr    e.g. "0071132"
+  //   [4]  Transaktions-Nr  e.g. "189945"
+  //   [5]  "KM PRODUKT"     e.g. "68040 DIESEL" (km and product merged, no tab between)
+  //   [6]  "CODE UNIT"      e.g. "0009 LTR"
+  //   [7]  Menge            e.g. "45,910"
+  //   [8]  Preis/E brutto   e.g. "2,1790"
+  //   [9]  Preis/E netto    e.g. "1,8312"
+  //   [10] Bezugswert brutto e.g. "100,04"
+  //   [11] Bezugswert netto  e.g. "84,07"
+  //   ...
+
+  private parseSingleTxLine(line: string): DkvTransaction | null {
+    const fields = line.split('\t').map(f => f.trim());
+    if (fields.length < 8) return null;
+
+    const datePattern = /^\d{2}\.\d{2}\.\d{4}$/;
+    if (!datePattern.test(fields[0])) return null;
+
+    // fields[5] = "KM_VALUE PRODUKT" — split on first whitespace boundary
+    const kmAndProdukt = fields[5].split(/\s+/);
+    const km = kmAndProdukt[0] ?? '0';
+    const produkt = kmAndProdukt.slice(1).join(' ');
+
+    // fields[6] = "UNITCODE EINHEIT" — Einheit is everything after the first token
+    const unitParts = fields[6].split(/\s+/);
+    const einheit = unitParts.slice(1).join(' ') || unitParts[0];
+
+    return {
+      lieferdatum: fields[0],
+      ort: fields[2] ?? '',
+      kilometerstand: parseDE(km),
+      produkt,
+      menge: parseDE(fields[7] ?? '0'),
+      einheit,
+      netto: parseDE(fields[11] ?? '0'),
+      brutto: parseDE(fields[10] ?? '0'),
+    };
+  }
+
+  // ─── Private: Multi-transaction columnar format ─────────────────────────────
+  //
+  // When a vehicle has multiple transactions, pdf-parse extracts the PDF table
+  // in columnar order (all dates, then all names, then all orts, etc.).
+  //
+  // Column groups (each group has n items, where n = transaction count):
+  //   group 0:  dates
+  //   group 1:  station names
+  //   group 2:  orts
+  //   group 3:  station numbers
+  //   group 4:  transaction numbers
+  //   group 5:  km values
+  //   group 6:  products
+  //   group 7:  unit codes (e.g. "0036")
+  //   group 8:  units (e.g. "LTR")
+  //   group 9:  quantities (Menge)
+  //   group 10: unit price brutto
+  //   group 11: unit price netto
+  //   group 12: total brutto (Bezugswert)
+  //   group 13: total netto  (Bezugswert)
+  //   ...       (discounts, taxes, final totals — not extracted)
+
+  private parseMultiTxColumnar(lines: string[]): DkvTransaction[] {
+    const datePattern = /^\d{2}\.\d{2}\.\d{4}$/;
+
+    // Count consecutive date lines at the start to determine n
+    let n = 0;
+    for (const line of lines) {
+      if (datePattern.test(line)) n++;
+      else break;
+    }
+    if (n === 0) return [];
+
+    const g = (groupIdx: number): string[] =>
+      lines.slice(groupIdx * n, (groupIdx + 1) * n);
+
+    const dates = g(0);
+    const orts = g(2);
+    const kms = g(5);
+    const products = g(6);
+    const units = g(8);
+    const quantities = g(9);
+    const totals_brutto = g(12);
+    const totals_netto = g(13);
+
+    return dates.map((date, i) => ({
+      lieferdatum: date,
+      ort: orts[i] ?? '',
+      // German number format: "19.234,56" → 19234.56 (strip dots, replace comma)
+      kilometerstand: parseDE(kms[i] ?? '0'),
+      produkt: products[i] ?? '',
+      menge: parseDE(quantities[i] ?? '0'),
+      einheit: units[i] ?? '',
+      netto: parseDE(totals_netto[i] ?? '0'),
+      brutto: parseDE(totals_brutto[i] ?? '0'),
+    }));
+  }
+}
+
+// ─── German number parser (module-level — used in private methods above) ──────
+function parseDE(raw: string): number {
+  return parseFloat(raw.replace(/\./g, '').replace(',', '.'));
+}
diff --git a/apps/api/src/dkv/dkv-parser.validate.ts b/apps/api/src/dkv/dkv-parser.validate.ts
new file mode 100644
index 0000000..b1220c9
--- /dev/null
+++ b/apps/api/src/dkv/dkv-parser.validate.ts
@@ -0,0 +1,207 @@
+/**
+ * Wave 0 validation script for the DKV PDF parser.
+ *
+ * Usage (from project root):
+ *   node --experimental-strip-types apps/api/src/dkv/dkv-parser.validate.ts
+ *
+ * Reads user-files/invoice.pdf, extracts text via pdf-parse v2,
+ * runs the vehicle-block regex and transaction parsers, prints results.
+ * Exits 1 if zero vehicle blocks are found (assertion failure).
+ */
+
+import { readFileSync } from 'node:fs';
+import { join } from 'node:path';
+import { PDFParse } from 'pdf-parse';
+import type { DkvVehicleBlock, DkvTransaction } from './dkv.types.js';
+
+// ─── German number parser ─────────────────────────────────────────────────────
+function parseDE(raw: string): number {
+  return parseFloat(raw.replace(/\./g, '').replace(',', '.'));
+}
+
+// ─── Single-transaction line parser ──────────────────────────────────────────
+// Format: DD.MM.YYYY \t Name \t Ort \t StNr \t TxNr \t "KM PRODUKT" \t "CODE UNIT" \t Menge \t PriceBrutto \t PriceNetto \t TotalBrutto \t TotalNetto \t ...
+function parseSingleTxLine(line: string): DkvTransaction | null {
+  const fields = line.split('\t').map(f => f.trim());
+  if (fields.length < 8) return null;
+
+  const datePattern = /^\d{2}\.\d{2}\.\d{4}$/;
+  if (!datePattern.test(fields[0])) return null;
+
+  // fields[5] = "KM_VALUE PRODUKT" (km and produkt merged — no tab between them in PDF)
+  const kmAndProdukt = fields[5].split(/\s+/);
+  const km = kmAndProdukt[0] ?? '0';
+  const produkt = kmAndProdukt.slice(1).join(' ');
+
+  // fields[6] = "UNITCODE UNIT" (e.g., "0009 LTR" or "0036 LTR")
+  const unitParts = fields[6].split(/\s+/);
+  const einheit = unitParts.slice(1).join(' ') || unitParts[0];
+
+  return {
+    lieferdatum: fields[0],
+    ort: fields[2] ?? '',
+    kilometerstand: parseDE(km),
+    produkt,
+    menge: parseDE(fields[7] ?? '0'),
+    einheit,
+    netto: parseDE(fields[11] ?? '0'),
+    brutto: parseDE(fields[10] ?? '0'),
+  };
+}
+
+// ─── Multi-transaction columnar parser ───────────────────────────────────────
+// Format (3 transactions with n=3):
+//   [0..2]    dates         n values
+//   [3..5]    names         n values
+//   [6..8]    orts          n values
+//   [9..11]   station_nrs   n values
+//   [12..14]  tx_nrs        n values
+//   [15..17]  km_values     n values
+//   [18..20]  products      n values
+//   [21..23]  unit_codes    n values
+//   [24..26]  units         n values
+//   [27..29]  quantities    n values
+//   [30..32]  unit_brutto   n values
+//   [33..35]  unit_netto    n values
+//   [36..38]  total_brutto  n values
+//   [39..41]  total_netto   n values
+//   ... (discounts, taxes, final totals)
+function parseMultiTxColumnar(lines: string[]): DkvTransaction[] {
+  // Count leading date lines to determine n (number of transactions)
+  const datePattern = /^\d{2}\.\d{2}\.\d{4}$/;
+  let n = 0;
+  for (const line of lines) {
+    if (datePattern.test(line)) n++;
+    else break;
+  }
+  if (n === 0) return [];
+
+  const g = (groupIdx: number): string[] =>
+    lines.slice(groupIdx * n, (groupIdx + 1) * n);
+
+  const dates = g(0);
+  const orts = g(2);
+  const kms = g(5);
+  const products = g(6);
+  const units = g(8);
+  const quantities = g(9);
+  const totals_brutto = g(12);
+  const totals_netto = g(13);
+
+  return dates.map((date, i) => ({
+    lieferdatum: date,
+    ort: orts[i] ?? '',
+    kilometerstand: parseDE(kms[i] ?? '0'),
+    produkt: products[i] ?? '',
+    menge: parseDE(quantities[i] ?? '0'),
+    einheit: units[i] ?? '',
+    netto: parseDE(totals_netto[i] ?? '0'),
+    brutto: parseDE(totals_brutto[i] ?? '0'),
+  }));
+}
+
+// ─── Transaction rows dispatcher ─────────────────────────────────────────────
+function parseTransactionRows(blockText: string): DkvTransaction[] {
+  const lines = blockText
+    .split('\n')
+    .map(l => l.trim())
+    .filter(l => l && !l.startsWith('»'));
+
+  if (lines.length === 0) return [];
+
+  const datePattern = /^\d{2}\.\d{2}\.\d{4}/;
+
+  // Detect format: if any date-starting line also contains a tab → single-tx format
+  const hasTabbedDateLine = lines.some(l => datePattern.test(l) && l.includes('\t'));
+
+  if (hasTabbedDateLine) {
+    // Single-transaction format
+    return lines
+      .filter(l => datePattern.test(l) && l.includes('\t'))
+      .map(parseSingleTxLine)
+      .filter((tx): tx is DkvTransaction => tx !== null);
+  }
+
+  // Multi-transaction columnar format
+  return parseMultiTxColumnar(lines);
+}
+
+// ─── Vehicle block parser ─────────────────────────────────────────────────────
+function parseDkvText(text: string): DkvVehicleBlock[] {
+  const vehicles: DkvVehicleBlock[] = [];
+
+  // Anchor on VEHICLE: marker — each block extends until next VEHICLE: or end of text
+  // Kennzeichen can contain letters, digits, hyphens, spaces, dots (e.g. "GP-JL 728E")
+  const vehicleBlockPattern =
+    /VEHICLE:\s+([A-Z0-9 ._\-]+?)\s+CARD NO\.:\s+(\S+)([\s\S]*?)(?=VEHICLE:|$)/g;
+
+  let match: RegExpExecArray | null;
+  while ((match = vehicleBlockPattern.exec(text)) !== null) {
+    const kennzeichen = match[1].trim();
+    const cardNumber = match[2].trim();
+    const blockText = match[3];
+
+    const transactions = parseTransactionRows(blockText);
+    vehicles.push({ kennzeichen, cardNumber, transactions });
+  }
+  return vehicles;
+}
+
+// ─── PDF extraction ───────────────────────────────────────────────────────────
+async function extractPdfText(buffer: Buffer): Promise<string> {
+  const parser = new PDFParse({ data: buffer });
+  const result = await parser.getText();
+  await parser.destroy(); // free memory per pdf-parse guidance (T-07-01)
+  return result.text;
+}
+
+// ─── Main ─────────────────────────────────────────────────────────────────────
+async function main(): Promise<void> {
+  const invoicePath = join(process.cwd(), 'user-files', 'invoice.pdf');
+  let buffer: Buffer;
+  try {
+    buffer = readFileSync(invoicePath);
+  } catch {
+    console.error(`ERROR: Cannot read invoice file at ${invoicePath}`);
+    process.exit(1);
+  }
+
+  console.log(`[DKV Validate] Reading ${invoicePath} (${buffer.length} bytes)`);
+
+  let text: string;
+  try {
+    text = await extractPdfText(buffer);
+  } catch (err) {
+    console.error('[DKV Validate] pdf-parse extraction failed:', err);
+    process.exit(1);
+  }
+
+  console.log(`[DKV Validate] Extracted ${text.length} characters of text`);
+
+  const vehicles = parseDkvText(text);
+
+  if (vehicles.length === 0) {
+    console.error('[DKV Validate] ASSERTION FAILED: Zero vehicle blocks parsed!');
+    console.error('[DKV Validate] Full extracted text for regex debugging:');
+    console.error(text);
+    process.exit(1);
+  }
+
+  const totalTx = vehicles.reduce((s, v) => s + v.transactions.length, 0);
+  console.log(`[DKV Validate] PASSED: ${vehicles.length} vehicle block(s), ${totalTx} transaction(s) total\n`);
+
+  // Print first 3 blocks
+  for (const [i, v] of vehicles.slice(0, 3).entries()) {
+    console.log(`Block ${i + 1}: ${v.kennzeichen} (card: ${v.cardNumber}) — ${v.transactions.length} tx`);
+    for (const tx of v.transactions.slice(0, 2)) {
+      console.log(`  ${tx.lieferdatum} @ ${tx.ort} | ${tx.kilometerstand} km | ${tx.menge} ${tx.einheit} ${tx.produkt}`);
+    }
+  }
+
+  console.log(`\nAll vehicles: ${vehicles.map(v => `${v.kennzeichen}(${v.transactions.length}tx)`).join(', ')}`);
+}
+
+main().catch(err => {
+  console.error('[DKV Validate] Unhandled error:', err);
+  process.exit(1);
+});