> ## Documentation Index
> Fetch the complete documentation index at: https://docs.techeval.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Compliance & Data Residency

> India-first infrastructure, DPDP Act alignment, and conversational latency benchmarks.

Enterprise technical screening requires strict adherence to data privacy laws, especially when handling candidate resumes, voice data, and webcam telemetry. TechEval.ai is built from the ground up for the Indian market, ensuring full compliance without sacrificing conversational speed.

## DPDP Act Alignment

The Digital Personal Data Protection (DPDP) Act mandates strict controls over how Indian citizens' data is processed, stored, and retained. Kovi aligns with these requirements natively:

* **100% Data Localization:** All compute (GCP Cloud Run) and database storage (Neon DB & Pinecone) are strictly provisioned in the **Mumbai (Asia-South1)** region. Candidate data is never routed through or stored on US or European servers.
* **Data Minimization:** Kovi extracts only the necessary technical parameters from parsed resumes. We do not store or process demographic data (age, gender, ethnicity) that could introduce bias into the AI evaluation.
* **Automated Expiration:** Interview transcripts and WebRTC telemetry are automatically purged upon customer request or at the end of the contracted retention lifecycle.

## The Latency Advantage

Routing standard AI voice agents through US-based servers introduces 800ms+ of latency, resulting in unnatural pauses and candidates accidentally talking over the AI.

By localizing our LangGraph nodes and WebRTC streaming in Mumbai, Kovi achieves conversational latency benchmarks designed for Indian broadband and 5G networks:

* **Tier 1 Cities (Bengaluru, Hyderabad, Mumbai, Delhi, Chennai):** Sub-400ms audio response time.
* **Tier 2 Cities:** Sub-600ms audio response time.

This localized infrastructure ensures Kovi feels like a seamless, natural phone call, eliminating the "walkie-talkie" delay common in basic OpenAI wrappers.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.