Enrichment Layers

Beyond core demographics, the library offers four progressive depth layers, designed for simulation, fairness audits and AI training data.

The 4 layers

Layer	Name	Description
1	Core Demographics	State, gender, religion, caste, names, languages, biological markers, address
2	Socio-Economic Outcomes	CIBIL credit score, health risk, literacy, employment vulnerability (configurable bias)
3	Narrative Documents	Loan applications, OPD records, Hinglish WhatsApp chats, school admissions, resumes, support chats
4	Agent Persona Prompts	LLM-ready system prompts, worldview beliefs, stress responses, memory seeds

Layer 2 — Outcomes

Simulated socio-economic outcomes calibrated to CMIE/CIBIL observed gaps.

# CLI
indian-fakedata -c 100 --outcomes

# Python
from indian_fakedata import generate_enriched

enriched = generate_enriched(count=10, include_outcomes=True, bias_level=0.3)[0]
credit = enriched["outcomes"]["credit"]["creditScore"]
health = enriched["outcomes"]["health"]["healthRisk"]

The bias dial

bias_level (CLI: --bias <0-1>) controls how strongly historical discrimination shapes outcomes:

Value	Meaning
`0.0`	Pure meritocracy
`0.3`	Default; calibrated to observed CMIE/CIBIL gaps
`1.0`	Max historical discrimination

Layer 3 — Narratives

Realistic Indian text documents generated from a profile:

Type	Document
`loan_application`	Bank loan application text
`medical_consultation`	Doctor's OPD consultation record
`school_enrollment`	School admission form narrative
`ration_card_application`	PDS ration card application
`hinglish_conversation`	WhatsApp-style Hinglish chat
`resume`	CV built from education timeline, work history and skills (v2.0.9)
`customer_support_chat`	Hinglish bank helpline dialogue, masked phone (v2.0.9)

# CLI
indian-fakedata -c 1000 --narrative loan_application --narrative hinglish_conversation

# Python
from indian_fakedata import generate, generate_narrative

profile = generate(count=1)[0]
doc = generate_narrative(profile, "loan_application")
print(doc["content"])

Layer 4 — Agent personas

LLM-ready agent personas: a short system prompt plus a complete fullPrompt roleplay prompt (identity, appearance, education timeline, personality traits, movie/anime preferences, habits, beliefs, memory seeds and behavior rules), ready to inject into any LLM. Since v2.0.9 personas accept a language — english (default), hindi (Devanagari prompt + Hindi headers) or hinglish (roman mix). CLI: --persona --persona-lang hindi.

# CLI
indian-fakedata -c 500 --persona -f jsonl -o agents.jsonl

# Python
from indian_fakedata import generate_persona

out = generate_persona(seed="011")
persona = out["persona"]

print(persona["systemPrompt"])     # ready to use as a chat system prompt
print(persona["fullPrompt"])       # complete self-contained roleplay prompt
print(persona["beliefs"]["worldview"])
print(persona["memorySeeds"])      # conversation-starting memories
print(persona["behaviorRules"])    # what the agent will and will not do

Everything at once

indian-fakedata -c 100 --enrich

--enrich = outcomes + all narrative types + persona.