Use Case Intelligence RSS Search use cases Search
NVIDIA Public Sector India

National Informatics Centre

Document Processing · Enterprise Knowledge Search

NIC needed to reduce judicial backlog and improve accessibility in a court system with massive case volumes and many official languages, while giving judges faster ways to search and summarize legal material.

Read the original story Explore all use cases
18,000 daily district court sessions
22 officially recognized languages
English and seven Indic languages
nearly 50,000 cases

Reported impact

Business outcome

NIC improved accessibility for multilingual court proceedings, accelerated legal research for judges, and created an AI platform that can support 18,000 daily district court sessions and scale to the wider Indian judicial system.

  • Customer Experience
  • Productivity Improvement
  • Scalability

Solution

NIC is modernizing India's court system with a two-part AI stack that handles both speech processing and legal retrieval. For court proceedings, streaming ASR models trained with NVIDIA NeMo and deployed with NVIDIA Riva transcribe English and seven Indic languages, while a Triton-hosted neural machine translation service extends access across all 22 Indic languages. For research workflows, Supreme Court judgments are embedded with the NVIDIA NeMo Embedding microservice, stored in a GPU-accelerated vector database, retrieved with NeMo Retriever, reranked, and passed to a TensorRT-LLM-optimized model served through NVIDIA NIM on NVIDIA DGX A100 infrastructure with NVIDIA AI Enterprise. The resulting platform supports transcription, translation, verdict search, and summarization, giving judges and court staff faster access to multilingual proceedings and relevant precedents while creating a base for system-wide rollout.

Data flow

Court audio is transcribed with Riva and NeMo-based ASR, translated through Triton-hosted NMT, and stored for access, while judgments are embedded, retrieved, reranked, and passed to an NIM-served TensorRT-LLM model to answer legal-search and summarization requests.

Solution architecture

10 components · 4 layers
  1. Ingest
    • NVIDIA NeMo Embedding microservice Converts Supreme Court judgments into embeddings for the judicial-search assistant.
  2. Compute
    • NVIDIA NeMo Trains the Indic language ASR models and supplies retrieval microservices for the legal-search stack.
    • NVIDIA DGX A100 Tensor Core GPUs Provide the accelerated infrastructure for the retrieval and language-model workloads.
  3. Serving
    • NVIDIA Riva Runs the streaming ASR service for court-proceeding transcription.
    • NVIDIA Triton Inference Server Hosts the neural machine translation service for multilingual court proceedings.
    • NVIDIA NeMo Retriever microservice Retrieves relevant verdict information from the vector database during question answering.
    • NeMo Reranking microservice Refines retrieved results before they are passed to the response model.
    • TensorRT-LLM Optimizes the large language model used in the judicial assistant.
    • NVIDIA NIM Serves the optimized LLM that generates answers from the retrieved legal context.
  4. Orchestration
    • NVIDIA AI Enterprise Provides the enterprise software stack used for embeddings, reranking, and inference.

Architecture clues

  • NIC also built extractive and abstractive summarization for pleadings
  • The RAG stack uses embeddings, retrieval, reranking, and a TensorRT-LLM-optimized model on DGX A100 GPUs
  • The translation service supports all 22 Indic languages while ASR currently supports eight languages including English

Evidence from the source

Results are refined with the NeMo Reranking microservice and fed into a TensorRT-LLM-optimized large language model deployed using LLM NIM.
The ASR/NMT solution will assist over 18,000 daily district court sessions after full rollout.
The neural machine translation (NMT) service supports the translation of court proceedings in all 22 Indic languages.
The streaming automatic speech recognition (ASR) service can transcribe speech in English and seven Indic languages.
Transparency Evidence & methodology 90/100 reference value · 89% confidence · 1 source

Review how this reference was scored and trace each normalized claim back to its documented source.

01

Scoring methodology

How scores are calculated

Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.

Customer reference value90/100
  • Problem clarity20/20
    Measures whether the customer's operational problem is specific enough to compare.
  • Implementation depth25/25
    Rewards named components, data flow, integrations, and architecture detail.
  • Outcome specificity20/20
    Measures whether the reported benefit is concrete rather than a generic claim.
  • Quantified impact15/25
    1 metric describe measurable impact.
  • Deployment scale10/10
    Rewards evidence of a live or scaled deployment and meaningful integration scope.
Evidence confidence89%
  • Source authority20/25
    The main evidence is an official vendor-published customer story.
  • Claim traceability21/25
    10 of 12 key fields link to a named source.
  • Direct evidence20/20
    4 direct source quotes retained.
  • Date provenance10/10
    The publication date comes from the source page or publisher.
  • Source transparency8/10
    1 source reference document publisher, URL, authority, retrieval, and supported fields.
  • Implementation specificity10/10
    Rewards source-backed components, data flow, quantified impact, and a specific outcome.
02

Traceability

Sources & provenance

Field coverage shows which facts each source supports. Dates are shown only when documented by the source.

  1. NVIDIA

    National Informatics Centre customer story

    First-party
    https://www.nvidia.com/en-us/case-studies/role-of-nic-modernizing-indian-judiciary-using-ai/
    Source type
    Vendor-published customer story
    Published
    Jan 17, 2025
    Supports
    • Customer
    • Business problem
    • Solution
    • Products
    • Solution components
    • Data flow
    • Integrations
    • technical area
    • use case category
    • Business outcome
    • outcome category
    • Metrics
    • Architecture
    • evidence quotes
    • Publication date

Date evidence

  • Published on source page Jan 17, 2025 · NVIDIA · First-party
    Open date source
Not separately sourced Industry, Region, Company size. Treat these normalized attributes as lower-confidence until a dedicated source is added.

Related

Similar use cases

See all like this
NVIDIA Jul 7, 2026 Public Ministry of Rio Grande do Sul (MPRS) Public Sector Brazil MPRS needed to manage rising legal caseloads, archived police inquiries, and public-service requests more efficiently, but the volume of legal documents and the need for compliant, locally controlled analysis made… 497 municipalities Document ProcessingEnterprise Knowledge Search Archiving Intelligence Assistant (AIA) · Citizen Access Agent · Justice Intelligence platform OpenAI Aug 31, 2026 Polimill Public Sector Asia-Pacific & Oceania Polimill needed to reduce administrative burden across fragmented Japanese municipal systems so public employees could find prior decisions, prepare assembly responses, and deliver services without each municipality… 3-5x development speed increase from Codex and OpenAI hands-on support Document ProcessingEnterprise Knowledge Search Codex · GPT models · OpenAI API Mistral May 25, 2026 Austrian Academy of Sciences Public Sector Austria Researchers at the Austrian Academy of Sciences needed a way to analyze and reconstruct huge volumes of fragmented Ancient Greek texts that manual transcription and interpretation could not process quickly or at… 1M+ unread Greek papyri worldwide, now accessible through AI Document ProcessingEnterprise Knowledge Search Apollo