diff --git a/.gitattributes b/.gitattributes
index 8e366e85..03cca427 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -6,3 +6,8 @@ model/quantized/tokenizer_config.json filter= diff= merge= text
 model/quantized/special_tokens_map.json filter= diff= merge= text
 model/quantized/added_tokens.json filter= diff= merge= text
 *.onnx filter=lfs diff=lfs merge=lfs -text
+
+# PII generator word pools are embedded via //go:embed and parsed line by line.
+# Pin them to LF so the bytes embedded into the binary are identical regardless
+# of the build platform's checkout settings.
+src/backend/pii/generators/data/*.txt text eol=lf
diff --git a/src/backend/pii/generators/data/cities.txt b/src/backend/pii/generators/data/cities.txt
new file mode 100644
index 00000000..a3b4da53
--- /dev/null
+++ b/src/backend/pii/generators/data/cities.txt
@@ -0,0 +1,74 @@
+# Dummy city pool for CityGenerator.
+# One value per line. Blank lines and lines starting with '#' are ignored.
+
+# US cities
+Springfield
+Riverside
+Greenville
+Fairview
+Madison
+Georgetown
+Salem
+Arlington
+Franklin
+Clinton
+Bristol
+Chester
+Dayton
+Kingston
+Newport
+Oakland
+Plymouth
+Burlington
+Manchester
+Lexington
+Milton
+Ashland
+Clayton
+Auburn
+Fairfield
+Marion
+Oxford
+Dover
+
+# Canadian cities
+Toronto
+Vancouver
+Calgary
+Ottawa
+Edmonton
+Winnipeg
+Halifax
+Victoria
+Regina
+Saskatoon
+Hamilton
+Kitchener
+London
+Windsor
+Burnaby
+Guelph
+
+# UK cities
+Birmingham
+Edinburgh
+Liverpool
+Leeds
+Sheffield
+Newcastle
+Nottingham
+Southampton
+Portsmouth
+Oxford
+Cambridge
+York
+Bath
+Brighton
+Cardiff
+Belfast
+Glasgow
+Aberdeen
+Dundee
+Swansea
+Norwich
+Exeter
diff --git a/src/backend/pii/generators/data/company_prefixes.txt b/src/backend/pii/generators/data/company_prefixes.txt
new file mode 100644
index 00000000..7ea6c8b9
--- /dev/null
+++ b/src/backend/pii/generators/data/company_prefixes.txt
@@ -0,0 +1,47 @@
+# Dummy company-name prefix pool for CompanyNameGenerator.
+# One value per line. Blank lines and lines starting with '#' are ignored.
+
+Acme
+Global
+United
+Pacific
+Atlantic
+Northern
+Summit
+Horizon
+Apex
+Vanguard
+Pinnacle
+Premier
+Elite
+Prime
+Sterling
+Meridian
+Coastal
+Central
+National
+Continental
+Metro
+Allied
+Dynamic
+Synergy
+Fusion
+Vertex
+Quantum
+Nova
+Titan
+Omega
+Pioneer
+Frontier
+Legacy
+Heritage
+Keystone
+Benchmark
+Catalyst
+Spectrum
+Nexus
+Compass
+Evergreen
+Cornerstone
+Beacon
+Cascade
diff --git a/src/backend/pii/generators/data/company_suffixes.txt b/src/backend/pii/generators/data/company_suffixes.txt
new file mode 100644
index 00000000..91bdb8c6
--- /dev/null
+++ b/src/backend/pii/generators/data/company_suffixes.txt
@@ -0,0 +1,39 @@
+# Dummy company-name suffix pool for CompanyNameGenerator.
+# One value per line. Blank lines and lines starting with '#' are ignored.
+
+# US/International
+Inc
+LLC
+Corp
+Industries
+Solutions
+Group
+Holdings
+Partners
+Enterprises
+Co
+Technologies
+Systems
+Services
+Consulting
+Associates
+International
+Worldwide
+Labs
+Ventures
+
+# UK
+Ltd
+PLC
+Limited
+
+# German
+GmbH
+AG
+
+# French/Spanish
+SA
+SL
+
+# Australian
+Pty Ltd
diff --git a/src/backend/pii/generators/data/countries.txt b/src/backend/pii/generators/data/countries.txt
new file mode 100644
index 00000000..6f1f7810
--- /dev/null
+++ b/src/backend/pii/generators/data/countries.txt
@@ -0,0 +1,78 @@
+# Dummy country pool for CountryGenerator.
+# One value per line. Blank lines and lines starting with '#' are ignored.
+
+# North America
+United States
+Canada
+Mexico
+
+# Europe
+United Kingdom
+Germany
+France
+Italy
+Spain
+Netherlands
+Belgium
+Switzerland
+Austria
+Sweden
+Norway
+Denmark
+Finland
+Ireland
+Poland
+Portugal
+Greece
+Czech Republic
+Hungary
+Romania
+Iceland
+Croatia
+
+# Asia
+Japan
+China
+South Korea
+India
+Singapore
+Thailand
+Vietnam
+Indonesia
+Malaysia
+Philippines
+Taiwan
+Hong Kong
+Nepal
+Sri Lanka
+
+# Oceania
+Australia
+New Zealand
+
+# South America
+Brazil
+Argentina
+Chile
+Colombia
+Peru
+Uruguay
+Ecuador
+
+# Africa
+South Africa
+Nigeria
+Kenya
+Egypt
+Morocco
+Ghana
+Tanzania
+Senegal
+
+# Middle East
+United Arab Emirates
+Israel
+Saudi Arabia
+Turkey
+Jordan
+Qatar
diff --git a/src/backend/pii/generators/data/email_firstnames.txt b/src/backend/pii/generators/data/email_firstnames.txt
new file mode 100644
index 00000000..1342860f
--- /dev/null
+++ b/src/backend/pii/generators/data/email_firstnames.txt
@@ -0,0 +1,78 @@
+# Dummy email local-part first names for EmailGenerator.
+# One value per line. Blank lines and lines starting with '#' are ignored.
+# Must stay lowercase single ASCII words (no spaces, apostrophes, or accents):
+# the local part is formatted as first.last and is checked against [a-z]+.
+
+# Western
+jane
+john
+alex
+sam
+taylor
+casey
+jordan
+riley
+michael
+sarah
+david
+emily
+james
+emma
+robert
+olivia
+joseph
+laura
+morgan
+avery
+
+# Asian
+wei
+mei
+hiroshi
+yuki
+jin
+min
+raj
+priya
+ravi
+deepa
+
+# African
+amara
+kofi
+zara
+kwame
+nia
+jelani
+thabo
+lerato
+
+# Middle Eastern
+yusuf
+fatima
+omar
+layla
+ali
+nadia
+samir
+rania
+
+# Latin American
+carlos
+maria
+diego
+sofia
+miguel
+lucia
+javier
+daniela
+
+# Eastern European
+dmitri
+anna
+ivan
+katya
+alexei
+elena
+boris
+zofia
diff --git a/src/backend/pii/generators/data/email_surnames.txt b/src/backend/pii/generators/data/email_surnames.txt
new file mode 100644
index 00000000..87e3a3e3
--- /dev/null
+++ b/src/backend/pii/generators/data/email_surnames.txt
@@ -0,0 +1,80 @@
+# Dummy email local-part surnames for EmailGenerator.
+# One value per line. Blank lines and lines starting with '#' are ignored.
+# Must stay lowercase single ASCII words (no spaces, apostrophes, or accents):
+# the local part is formatted as first.last and is checked against [a-z]+.
+
+# Western
+doe
+smith
+johnson
+brown
+davis
+wilson
+moore
+taylor
+anderson
+thomas
+jackson
+white
+harris
+martin
+thompson
+clark
+lewis
+walker
+young
+
+# Asian
+chen
+wang
+kim
+nguyen
+tanaka
+yamamoto
+patel
+singh
+kumar
+sharma
+
+# African
+okonkwo
+diallo
+mensah
+osei
+abebe
+adeyemi
+owusu
+
+# Middle Eastern
+mohammed
+ahmed
+hassan
+khan
+ali
+ibrahim
+malik
+
+# Latin American
+garcia
+rodriguez
+martinez
+lopez
+gonzalez
+hernandez
+perez
+torres
+
+# Eastern European
+ivanov
+petrov
+kowalski
+novak
+horvat
+popov
+
+# Celtic
+obrien
+murphy
+kelly
+sullivan
+walsh
diff --git a/src/backend/pii/generators/data/firstnames.txt b/src/backend/pii/generators/data/firstnames.txt
new file mode 100644
index 00000000..56e0c305
--- /dev/null
+++ b/src/backend/pii/generators/data/firstnames.txt
@@ -0,0 +1,145 @@
+# Dummy first-name pool for FirstNameGenerator.
+# One value per line. Blank lines and lines starting with '#' are ignored.
+# Grouped by rough region only for readability; order does not affect output.
+
+# Western
+John
+Jane
+Michael
+Sarah
+David
+Emily
+James
+Emma
+Robert
+Olivia
+William
+Elizabeth
+Richard
+Jennifer
+Thomas
+Jessica
+Charles
+Amanda
+Christopher
+Ashley
+Daniel
+Stephanie
+Matthew
+Nicole
+Anthony
+Melissa
+Joseph
+Margaret
+Andrew
+Laura
+Benjamin
+Hannah
+Samuel
+Grace
+Nathan
+Rebecca
+
+# Asian
+Wei
+Mei
+Hiroshi
+Yuki
+Jin
+Min
+Raj
+Priya
+Kenji
+Sakura
+Chen
+Li
+Aiko
+Takeshi
+Ananya
+Arjun
+Haruto
+Yui
+Ravi
+Deepa
+Jae
+Soo
+
+# African
+Amara
+Kofi
+Zara
+Kwame
+Nia
+Jelani
+Amina
+Chioma
+Oluwaseun
+Aisha
+Tariq
+Imani
+Sekou
+Adaeze
+Kwabena
+Ama
+Thabo
+Lerato
+
+# Middle Eastern
+Yusuf
+Fatima
+Omar
+Layla
+Ali
+Nadia
+Hassan
+Mariam
+Khalid
+Zahra
+Ahmed
+Leila
+Ibrahim
+Yasmin
+Samir
+Rania
+Karim
+Salma
+
+# Latin American
+Carlos
+Maria
+Diego
+Sofia
+Miguel
+Lucia
+Alejandro
+Valentina
+Fernando
+Camila
+Ricardo
+Isabella
+Andres
+Gabriela
+Javier
+Daniela
+Mateo
+Ximena
+
+# Eastern European
+Dmitri
+Anna
+Ivan
+Katya
+Alexei
+Elena
+Nikolai
+Olga
+Sergei
+Natasha
+Vladimir
+Irina
+Mikhail
+Tatiana
+Boris
+Zofia
+Pavel
+Ludmila
diff --git a/src/backend/pii/generators/data/states.txt b/src/backend/pii/generators/data/states.txt
new file mode 100644
index 00000000..5193a319
--- /dev/null
+++ b/src/backend/pii/generators/data/states.txt
@@ -0,0 +1,90 @@
+# Dummy state / region pool for StateGenerator.
+# One value per line. Blank lines and lines starting with '#' are ignored.
+
+# US States
+Alabama
+Alaska
+Arizona
+Arkansas
+California
+Colorado
+Connecticut
+Delaware
+Florida
+Georgia
+Hawaii
+Idaho
+Illinois
+Indiana
+Iowa
+Kansas
+Kentucky
+Louisiana
+Maine
+Maryland
+Massachusetts
+Michigan
+Minnesota
+Mississippi
+Missouri
+Montana
+Nebraska
+Nevada
+New Hampshire
+New Jersey
+New Mexico
+New York
+North Carolina
+North Dakota
+Ohio
+Oklahoma
+Oregon
+Pennsylvania
+Rhode Island
+South Carolina
+South Dakota
+Tennessee
+Texas
+Utah
+Vermont
+Virginia
+Washington
+West Virginia
+Wisconsin
+Wyoming
+
+# Canadian Provinces
+Ontario
+Quebec
+British Columbia
+Alberta
+Manitoba
+Saskatchewan
+Nova Scotia
+New Brunswick
+Newfoundland and Labrador
+Prince Edward Island
+
+# UK Regions/Nations
+England
+Scotland
+Wales
+Northern Ireland
+
+# UK Counties
+Yorkshire
+Kent
+Essex
+Hampshire
+Surrey
+Lancashire
+Devon
+Cornwall
+Oxfordshire
+Cambridgeshire
+Berkshire
+Somerset
+Dorset
+Wiltshire
+Norfolk
+Suffolk
diff --git a/src/backend/pii/generators/data/streets.txt b/src/backend/pii/generators/data/streets.txt
new file mode 100644
index 00000000..aab35587
--- /dev/null
+++ b/src/backend/pii/generators/data/streets.txt
@@ -0,0 +1,54 @@
+# Dummy street-name pool for StreetGenerator.
+# One value per line. Blank lines and lines starting with '#' are ignored.
+
+# US style streets
+Main St
+Oak Ave
+Maple Dr
+Park Blvd
+Elm Street
+Pine Road
+Cedar Lane
+Washington St
+Broadway
+Market St
+Church St
+Mill Road
+School Lane
+Lake Ave
+River Road
+Highland Ave
+Forest Dr
+Valley Road
+Sunset Blvd
+Spring St
+Garden Way
+Lincoln Ave
+Jefferson St
+Franklin Blvd
+Madison Ave
+Monroe Dr
+Adams St
+Willow Way
+Birch Lane
+Hillcrest Dr
+
+# UK/Canada style streets
+High Street
+Station Road
+Church Lane
+Victoria Road
+Queens Road
+King Street
+Manor Road
+Park Lane
+The Crescent
+Green Lane
+Mill Lane
+New Road
+Chapel Street
+West End
+North Terrace
+Kingsway
+Meadow Lane
+Orchard Close
diff --git a/src/backend/pii/generators/data/surnames.txt b/src/backend/pii/generators/data/surnames.txt
new file mode 100644
index 00000000..0394a478
--- /dev/null
+++ b/src/backend/pii/generators/data/surnames.txt
@@ -0,0 +1,156 @@
+# Dummy surname pool for SurnameGenerator.
+# One value per line. Blank lines and lines starting with '#' are ignored.
+
+# Western
+Smith
+Johnson
+Williams
+Brown
+Jones
+Garcia
+Miller
+Davis
+Martinez
+Wilson
+Anderson
+Taylor
+Thomas
+Moore
+Jackson
+Martin
+Lee
+Thompson
+White
+Harris
+Clark
+Lewis
+Robinson
+Walker
+Hall
+Young
+King
+Wright
+Hill
+Scott
+Green
+Baker
+Adams
+Nelson
+Carter
+Mitchell
+
+# Asian
+Chen
+Wang
+Li
+Zhang
+Liu
+Kim
+Park
+Choi
+Nguyen
+Tran
+Tanaka
+Yamamoto
+Suzuki
+Watanabe
+Patel
+Singh
+Sharma
+Kumar
+Yang
+Huang
+Sato
+Ito
+
+# African
+Okonkwo
+Diallo
+Mensah
+Osei
+Abebe
+Adeyemi
+Nkosi
+Mbeki
+Kamara
+Toure
+Dlamini
+Ndlovu
+Owusu
+Achebe
+Baldé
+Mwangi
+
+# Middle Eastern
+Mohammed
+Ahmed
+Hassan
+Khan
+Ali
+Ibrahim
+Hussein
+Malik
+Nazari
+Hosseini
+Rahman
+Begum
+Saleh
+Karimi
+Farah
+Aziz
+
+# Latin American
+Rodriguez
+Lopez
+Gonzalez
+Hernandez
+Perez
+Sanchez
+Ramirez
+Torres
+Flores
+Rivera
+Gomez
+Diaz
+Reyes
+Morales
+Cruz
+Ortiz
+Castillo
+Vargas
+
+# Eastern European
+Ivanov
+Petrov
+Kowalski
+Novak
+Horvat
+Popov
+Volkov
+Kozlov
+Nowak
+Kovalenko
+Bondarenko
+Shevchenko
+Sokolov
+Wojcik
+Kucera
+Marek
+
+# Celtic
+O'Brien
+Murphy
+Kelly
+Sullivan
+O'Connor
+Walsh
+Ryan
+Byrne
+MacDonald
+Campbell
+Stewart
+Murray
+Fraser
+MacLeod
+Doyle
+Gallagher
diff --git a/src/backend/pii/generators/pii_generators.go b/src/backend/pii/generators/pii_generators.go
index 08ab79e1..5154d729 100644
--- a/src/backend/pii/generators/pii_generators.go
+++ b/src/backend/pii/generators/pii_generators.go
@@ -29,43 +29,11 @@ func pickExcluding(rng *rand.Rand, choices []string, original string) string {
 
 // EmailGenerator generates dummy email addresses
 func EmailGenerator(rng *rand.Rand, original string) string {
-	firstNames := []string{
-		// Western names
-		"jane", "john", "alex", "sam", "taylor", "casey", "jordan", "riley",
-		"michael", "sarah", "david", "emily", "james", "emma", "robert", "olivia",
-		// Asian names
-		"wei", "mei", "hiroshi", "yuki", "jin", "min", "raj", "priya",
-		// African names
-		"amara", "kofi", "zara", "kwame", "nia", "jelani",
-		// Middle Eastern names
-		"yusuf", "fatima", "omar", "layla", "ali", "nadia",
-		// Latin American names
-		"carlos", "maria", "diego", "sofia", "miguel", "lucia",
-		// Eastern European names
-		"dmitri", "anna", "ivan", "katya", "alexei", "elena",
-	}
-	lastNames := []string{
-		// Western surnames
-		"doe", "smith", "johnson", "brown", "davis", "wilson", "moore", "taylor",
-		"anderson", "thomas", "jackson", "white", "harris", "martin", "thompson",
-		// Asian surnames
-		"chen", "wang", "kim", "nguyen", "tanaka", "yamamoto", "patel", "singh",
-		// African surnames
-		"okonkwo", "diallo", "mensah", "osei", "abebe",
-		// Middle Eastern surnames
-		"mohammed", "ahmed", "hassan", "khan", "ali",
-		// Latin American surnames
-		"garcia", "rodriguez", "martinez", "lopez", "gonzalez", "hernandez",
-		// Eastern European surnames
-		"ivanov", "petrov", "kowalski", "novak", "horvat",
-		// Celtic surnames
-		"obrien", "murphy", "kelly", "sullivan",
-	}
 	// RFC 2606 / RFC 6761 reserved domains only
 	domains := []string{"example.com", "example.org", "example.net", "test.com", "test.org", "test.net", "invalid.com", "invalid.org"}
 
-	firstName := firstNames[rng.Intn(len(firstNames))]
-	lastName := lastNames[rng.Intn(len(lastNames))]
+	firstName := emailFirstNames[rng.Intn(len(emailFirstNames))]
+	lastName := emailSurnames[rng.Intn(len(emailSurnames))]
 	domain := domains[rng.Intn(len(domains))]
 
 	return fmt.Sprintf("%s.%s@%s", firstName, lastName, domain)
@@ -181,18 +149,7 @@ func DateOfBirthGenerator(rng *rand.Rand, original string) string {
 
 // StreetGenerator generates dummy street addresses
 func StreetGenerator(rng *rand.Rand, original string) string {
-	streetNames := []string{
-		// US style streets
-		"Main St", "Oak Ave", "Maple Dr", "Park Blvd", "Elm Street", "Pine Road", "Cedar Lane", "Washington St",
-		"Broadway", "Market St", "Church St", "Mill Road", "School Lane", "Lake Ave", "River Road",
-		"Highland Ave", "Forest Dr", "Valley Road", "Sunset Blvd", "Spring St", "Garden Way",
-		"Lincoln Ave", "Jefferson St", "Franklin Blvd", "Madison Ave", "Monroe Dr",
-		// UK/Canada style streets
-		"High Street", "Station Road", "Church Lane", "Victoria Road", "Queens Road",
-		"King Street", "Manor Road", "Park Lane", "The Crescent", "Green Lane",
-		"Mill Lane", "New Road", "Chapel Street", "West End", "North Terrace",
-	}
-	return pickExcluding(rng, streetNames, original)
+	return pickExcluding(rng, streets, original)
 }
 
 // ZipCodeGenerator generates dummy zip codes
@@ -211,19 +168,6 @@ func ZipCodeGenerator(rng *rand.Rand, original string) string {
 
 // CityGenerator generates dummy city names
 func CityGenerator(rng *rand.Rand, original string) string {
-	cities := []string{
-		// US cities
-		"Springfield", "Riverside", "Greenville", "Fairview", "Madison", "Georgetown", "Salem", "Arlington",
-		"Franklin", "Clinton", "Bristol", "Chester", "Dayton", "Kingston", "Newport", "Oakland",
-		"Plymouth", "Burlington", "Manchester", "Lexington", "Milton", "Ashland", "Clayton",
-		// Canadian cities
-		"Toronto", "Vancouver", "Calgary", "Ottawa", "Edmonton", "Winnipeg", "Halifax",
-		"Victoria", "Regina", "Saskatoon", "Hamilton", "Kitchener", "London", "Windsor",
-		// UK cities
-		"Birmingham", "Edinburgh", "Liverpool", "Leeds", "Sheffield", "Newcastle",
-		"Nottingham", "Southampton", "Portsmouth", "Oxford", "Cambridge", "York", "Bath",
-		"Brighton", "Cardiff", "Belfast", "Glasgow", "Aberdeen", "Dundee", "Swansea",
-	}
 	return pickExcluding(rng, cities, original)
 }
 
@@ -242,56 +186,11 @@ func BuildingNumGenerator(rng *rand.Rand, original string) string {
 
 // FirstNameGenerator generates dummy first names
 func FirstNameGenerator(rng *rand.Rand, original string) string {
-	names := []string{
-		// Western names
-		"John", "Jane", "Michael", "Sarah", "David", "Emily", "James", "Emma", "Robert", "Olivia",
-		"William", "Elizabeth", "Richard", "Jennifer", "Thomas", "Jessica", "Charles", "Amanda",
-		"Christopher", "Ashley", "Daniel", "Stephanie", "Matthew", "Nicole", "Anthony", "Melissa",
-		// Asian names
-		"Wei", "Mei", "Hiroshi", "Yuki", "Jin", "Min", "Raj", "Priya",
-		"Kenji", "Sakura", "Chen", "Li", "Aiko", "Takeshi", "Ananya", "Arjun",
-		// African names
-		"Amara", "Kofi", "Zara", "Kwame", "Nia", "Jelani", "Amina", "Chioma",
-		"Oluwaseun", "Aisha", "Tariq", "Imani", "Sekou", "Adaeze",
-		// Middle Eastern names
-		"Yusuf", "Fatima", "Omar", "Layla", "Ali", "Nadia", "Hassan", "Mariam",
-		"Khalid", "Zahra", "Ahmed", "Leila", "Ibrahim", "Yasmin",
-		// Latin American names
-		"Carlos", "Maria", "Diego", "Sofia", "Miguel", "Lucia", "Alejandro", "Valentina",
-		"Fernando", "Camila", "Ricardo", "Isabella", "Andres", "Gabriela",
-		// Eastern European names
-		"Dmitri", "Anna", "Ivan", "Katya", "Alexei", "Elena", "Nikolai", "Olga",
-		"Sergei", "Natasha", "Vladimir", "Irina", "Mikhail", "Tatiana",
-	}
-	return pickExcluding(rng, names, original)
+	return pickExcluding(rng, firstNames, original)
 }
 
 // SurnameGenerator generates dummy last names
 func SurnameGenerator(rng *rand.Rand, original string) string {
-	surnames := []string{
-		// Western surnames
-		"Smith", "Johnson", "Williams", "Brown", "Jones", "Garcia", "Miller", "Davis", "Martinez", "Wilson",
-		"Anderson", "Taylor", "Thomas", "Moore", "Jackson", "Martin", "Lee", "Thompson", "White", "Harris",
-		"Clark", "Lewis", "Robinson", "Walker", "Hall", "Young", "King", "Wright", "Hill", "Scott",
-		// Asian surnames
-		"Chen", "Wang", "Li", "Zhang", "Liu", "Kim", "Park", "Choi", "Nguyen", "Tran",
-		"Tanaka", "Yamamoto", "Suzuki", "Watanabe", "Patel", "Singh", "Sharma", "Kumar",
-		// African surnames
-		"Okonkwo", "Diallo", "Mensah", "Osei", "Abebe", "Adeyemi", "Nkosi", "Mbeki",
-		"Kamara", "Toure", "Dlamini", "Ndlovu",
-		// Middle Eastern surnames
-		"Mohammed", "Ahmed", "Hassan", "Khan", "Ali", "Ibrahim", "Hussein", "Malik",
-		"Nazari", "Hosseini", "Rahman", "Begum",
-		// Latin American surnames
-		"Rodriguez", "Lopez", "Gonzalez", "Hernandez", "Perez", "Sanchez", "Ramirez", "Torres",
-		"Flores", "Rivera", "Gomez", "Diaz", "Reyes", "Morales", "Cruz", "Ortiz",
-		// Eastern European surnames
-		"Ivanov", "Petrov", "Kowalski", "Novak", "Horvat", "Popov", "Volkov", "Kozlov",
-		"Nowak", "Kovalenko", "Bondarenko", "Shevchenko",
-		// Celtic surnames
-		"O'Brien", "Murphy", "Kelly", "Sullivan", "O'Connor", "Walsh", "Ryan", "Byrne",
-		"MacDonald", "Campbell", "Stewart", "Murray", "Fraser", "MacLeod",
-	}
 	return pickExcluding(rng, surnames, original)
 }
 
@@ -342,76 +241,19 @@ func UrlGenerator(rng *rand.Rand, original string) string {
 
 // CompanyNameGenerator generates dummy company names
 func CompanyNameGenerator(rng *rand.Rand, original string) string {
-	prefixes := []string{
-		"Acme", "Global", "United", "Pacific", "Atlantic", "Northern", "Summit", "Horizon", "Apex", "Vanguard",
-		"Pinnacle", "Premier", "Elite", "Prime", "Sterling", "Meridian", "Coastal", "Central", "National", "Continental",
-		"Metro", "Allied", "Dynamic", "Synergy", "Fusion", "Vertex", "Quantum", "Nova", "Titan", "Omega",
-		"Pioneer", "Frontier", "Legacy", "Heritage", "Keystone", "Benchmark", "Catalyst", "Spectrum", "Nexus", "Compass",
-	}
-	suffixes := []string{
-		// US/International
-		"Inc", "LLC", "Corp", "Industries", "Solutions", "Group", "Holdings", "Partners", "Enterprises", "Co",
-		"Technologies", "Systems", "Services", "Consulting", "Associates", "International", "Worldwide",
-		// UK
-		"Ltd", "PLC", "Limited",
-		// German
-		"GmbH", "AG",
-		// French/Spanish
-		"SA", "SL",
-		// Australian
-		"Pty Ltd",
-	}
-
-	prefix := prefixes[rng.Intn(len(prefixes))]
-	suffix := suffixes[rng.Intn(len(suffixes))]
+	prefix := companyPrefixes[rng.Intn(len(companyPrefixes))]
+	suffix := companySuffixes[rng.Intn(len(companySuffixes))]
 
 	return fmt.Sprintf("%s %s", prefix, suffix)
 }
 
 // StateGenerator generates dummy state names
 func StateGenerator(rng *rand.Rand, original string) string {
-	states := []string{
-		// US States
-		"Alabama", "Alaska", "Arizona", "Arkansas", "California", "Colorado", "Connecticut", "Delaware",
-		"Florida", "Georgia", "Hawaii", "Idaho", "Illinois", "Indiana", "Iowa", "Kansas",
-		"Kentucky", "Louisiana", "Maine", "Maryland", "Massachusetts", "Michigan", "Minnesota", "Mississippi",
-		"Missouri", "Montana", "Nebraska", "Nevada", "New Hampshire", "New Jersey", "New Mexico", "New York",
-		"North Carolina", "North Dakota", "Ohio", "Oklahoma", "Oregon", "Pennsylvania", "Rhode Island", "South Carolina",
-		"South Dakota", "Tennessee", "Texas", "Utah", "Vermont", "Virginia", "Washington", "West Virginia",
-		"Wisconsin", "Wyoming",
-		// Canadian Provinces
-		"Ontario", "Quebec", "British Columbia", "Alberta", "Manitoba", "Saskatchewan",
-		"Nova Scotia", "New Brunswick", "Newfoundland and Labrador", "Prince Edward Island",
-		// UK Regions/Nations
-		"England", "Scotland", "Wales", "Northern Ireland",
-		// UK Counties
-		"Yorkshire", "Kent", "Essex", "Hampshire", "Surrey", "Lancashire", "Devon", "Cornwall",
-		"Oxfordshire", "Cambridgeshire", "Berkshire", "Somerset", "Dorset", "Wiltshire", "Norfolk", "Suffolk",
-	}
 	return pickExcluding(rng, states, original)
 }
 
 // CountryGenerator generates dummy country names
 func CountryGenerator(rng *rand.Rand, original string) string {
-	countries := []string{
-		// North America
-		"United States", "Canada", "Mexico",
-		// Europe
-		"United Kingdom", "Germany", "France", "Italy", "Spain", "Netherlands", "Belgium",
-		"Switzerland", "Austria", "Sweden", "Norway", "Denmark", "Finland", "Ireland",
-		"Poland", "Portugal", "Greece", "Czech Republic", "Hungary", "Romania",
-		// Asia
-		"Japan", "China", "South Korea", "India", "Singapore", "Thailand", "Vietnam",
-		"Indonesia", "Malaysia", "Philippines", "Taiwan", "Hong Kong",
-		// Oceania
-		"Australia", "New Zealand",
-		// South America
-		"Brazil", "Argentina", "Chile", "Colombia", "Peru",
-		// Africa
-		"South Africa", "Nigeria", "Kenya", "Egypt", "Morocco", "Ghana",
-		// Middle East
-		"United Arab Emirates", "Israel", "Saudi Arabia", "Turkey",
-	}
 	return pickExcluding(rng, countries, original)
 }
 
diff --git a/src/backend/pii/generators/pools.go b/src/backend/pii/generators/pools.go
new file mode 100644
index 00000000..edade87f
--- /dev/null
+++ b/src/backend/pii/generators/pools.go
@@ -0,0 +1,69 @@
+package pii
+
+import (
+	"bufio"
+	"embed"
+	"fmt"
+	"io"
+	"strings"
+)
+
+// poolData holds the newline-delimited word pools the generators draw from.
+// Keeping them as embedded data files rather than inline slices lets the pools
+// grow (and gain locale-aware variants later) without touching Go code, while
+// the binary stays self-contained.
+//
+//go:embed data/*.txt
+var poolData embed.FS
+
+// parsePool reads a newline-delimited pool from r. Surrounding whitespace is
+// trimmed, and blank lines and comment lines (those beginning with '#') are
+// skipped so the data files can carry headers and section labels.
+func parsePool(r io.Reader) []string {
+	var entries []string
+	scanner := bufio.NewScanner(r)
+	for scanner.Scan() {
+		line := strings.TrimSpace(scanner.Text())
+		if line == "" || strings.HasPrefix(line, "#") {
+			continue
+		}
+		entries = append(entries, line)
+	}
+	if err := scanner.Err(); err != nil {
+		panic(fmt.Sprintf("pii generators: reading pool: %v", err))
+	}
+	return entries
+}
+
+// loadPool loads a pool file from the embedded data/ directory. It panics on a
+// missing, unreadable, or empty file: the files are embedded at compile time,
+// so any of those is a build or packaging error that should fail loudly at
+// startup rather than leave a generator with an empty pool.
+func loadPool(name string) []string {
+	f, err := poolData.Open("data/" + name)
+	if err != nil {
+		panic(fmt.Sprintf("pii generators: cannot open embedded pool %q: %v", name, err))
+	}
+	defer f.Close()
+
+	entries := parsePool(f)
+	if len(entries) == 0 {
+		panic(fmt.Sprintf("pii generators: embedded pool %q is empty", name))
+	}
+	return entries
+}
+
+// Word pools loaded once at package initialization from the embedded data
+// files. The generators reference these directly.
+var (
+	emailFirstNames = loadPool("email_firstnames.txt")
+	emailSurnames   = loadPool("email_surnames.txt")
+	firstNames      = loadPool("firstnames.txt")
+	surnames        = loadPool("surnames.txt")
+	cities          = loadPool("cities.txt")
+	streets         = loadPool("streets.txt")
+	states          = loadPool("states.txt")
+	countries       = loadPool("countries.txt")
+	companyPrefixes = loadPool("company_prefixes.txt")
+	companySuffixes = loadPool("company_suffixes.txt")
+)
diff --git a/src/backend/pii/generators/pools_test.go b/src/backend/pii/generators/pools_test.go
new file mode 100644
index 00000000..6ca7ce20
--- /dev/null
+++ b/src/backend/pii/generators/pools_test.go
@@ -0,0 +1,81 @@
+package pii
+
+import (
+	"regexp"
+	"strings"
+	"testing"
+)
+
+func TestParsePool(t *testing.T) {
+	input := strings.Join([]string{
+		"# a comment",
+		"",
+		"  Alpha  ",
+		"Beta",
+		"   # indented comment",
+		"\tGamma\t",
+		"   ",
+		"Delta",
+	}, "\n")
+
+	got := parsePool(strings.NewReader(input))
+	want := []string{"Alpha", "Beta", "Gamma", "Delta"}
+
+	if len(got) != len(want) {
+		t.Fatalf("parsePool returned %d entries (%v), want %d (%v)", len(got), got, len(want), want)
+	}
+	for i := range want {
+		if got[i] != want[i] {
+			t.Errorf("parsePool entry %d = %q, want %q", i, got[i], want[i])
+		}
+	}
+}
+
+// loadedPools is every pool the generators draw from, so the checks below cover
+// them uniformly.
+var loadedPools = map[string][]string{
+	"email_firstnames": emailFirstNames,
+	"email_surnames":   emailSurnames,
+	"firstnames":       firstNames,
+	"surnames":         surnames,
+	"cities":           cities,
+	"streets":          streets,
+	"states":           states,
+	"countries":        countries,
+	"company_prefixes": companyPrefixes,
+	"company_suffixes": companySuffixes,
+}
+
+// Every embedded pool must load with real entries and no parser leakage
+// (blank lines or comment markers surviving into the data).
+func TestLoadedPoolsAreCleanAndNonEmpty(t *testing.T) {
+	for name, pool := range loadedPools {
+		if len(pool) == 0 {
+			t.Errorf("pool %q loaded empty", name)
+		}
+		for _, entry := range pool {
+			if entry == "" {
+				t.Errorf("pool %q contains an empty entry", name)
+			}
+			if strings.HasPrefix(entry, "#") {
+				t.Errorf("pool %q contains an unstripped comment: %q", name, entry)
+			}
+			if entry != strings.TrimSpace(entry) {
+				t.Errorf("pool %q entry has surrounding whitespace: %q", name, entry)
+			}
+		}
+	}
+}
+
+// EmailGenerator formats the local part as first.last and callers assert it
+// matches [a-z]+\.[a-z]+, so the email pools must stay lowercase ASCII words.
+func TestEmailPoolsAreLowercaseTokens(t *testing.T) {
+	tokenPattern := regexp.MustCompile(`^[a-z]+$`)
+	for _, name := range []string{"email_firstnames", "email_surnames"} {
+		for _, entry := range loadedPools[name] {
+			if !tokenPattern.MatchString(entry) {
+				t.Errorf("pool %q entry %q is not a lowercase ASCII token", name, entry)
+			}
+		}
+	}
+}