Multilingual Translation Administration

Governance portal for 22 Scheduled Indian Languages + English across LifeEasy Staffing Company.

Total Supported Locales

100%

Dictionary Coverage Rate

RTL Layouts (Urdu, Kashmiri, Sindhi)

Protected Statutory Terms

23 Languages Deployment Matrix

Locale	Language Name	Native Name	Script	Direction	IndicTrans2 Code	Status	Action

Side-by-Side Content Review & Editor

Target Locale: Content Key:

English Source (Default Reference)

Target Language Translation (Marathi)

Status: Approved

HR & Staffing Terminology Glossary

Enforces consistent statutory terminology across all 23 languages

The following statutory abbreviations, brand names, and HR terms are strictly preserved and verified to maintain legal compliance under the Central Labour Codes, EPF Act, ESI Act, and DPDP Act.

AI4Bharat IndicTrans2 Architecture & VPS Requirements

Primary Model Architecture: ai4bharat/indictrans2-en-indic-1B / ai4bharat/indictrans2-indic-en-1B

Repository: https://github.com/AI4Bharat/IndicTrans2

Hardware & VPS Specifications

Component	Inference Requirement	Static Delivery (Hostinger VPS)
CPU	8 vCPUs (Intel Xeon / AMD EPYC)	2 vCPUs sufficient for static caching
RAM	16 GB RAM minimum	4 GB RAM sufficient
GPU	NVIDIA T4 / A10G (16GB VRAM recommended)	No GPU required for static cached bundles
Model Weight Size	4.2 GB (FP16 Quantized)	0 MB (Served via pre-compiled JSON)
Throughput	~120 tokens/sec on GPU	Instant sub-5ms client rendering from cache

Architectural Note: As per production best practices, model inference is conducted asynchronously in batch pipelines. The public-facing website delivers compiled, high-performance static translation bundles (locales-data.js and /locales/*.json) avoiding heavy GPU overhead on standard web hosting tiers.