---
title: "Vinay M H - Lead Associate (AI) in Bengaluru | Tal"
description: "Meet Vinay M H, Multilingual Voice AI & LLM Engineer | RAG · Agentic AI · LLMOps | IIT Madras based in Bengaluru. AI stack includes claude code, claude, and codex. Explore public work links including github."
image: "https://tal.club/meet/vinay-m-h/share-image"
---

# Vinay M H on Tal

## Headline

Multilingual Voice AI & LLM Engineer \| RAG · Agentic AI · LLMOps \| IIT Madras

## About

I build AI systems that run at scale — voice agents processing millions of calls, RAG pipelines answering policy questions in four Indian languages, and LLM workflows that actually ship to production\. At I\-PAC, I architected a multilingual voice AI survey system \(English, Hindi, Tamil, Telugu\) integrating Vapi, ElevenLabs, Whisper, and Sarvam LLM — live across 20 lakh\+ telephony calls with real\-time transcription and structured JSON extraction\. I also built Scheme\.ai, a production RAG chatbot over government schemes and news using GPT\-4\.1, Groq Llama\-3, ChromaDB, and pgvector for sub\-second multilingual policy QA\. Before that, at VerifiedTalent, I cut OpenAI API latency by 40% through caching and prompt optimization, and engineered an LLM \+ NER profile enrichment system hitting 100% schema\-conformant data ingestion\. At OpenDataFabric, I deployed Llama\-2 on AWS SageMaker and built the backend infrastructure on GCP that lifted customer interaction efficiency by 25%\. What I work with: GPT\-4\.1, Claude, Llama\-3, Qwen3, Mistral · vLLM, Groq · LangChain, LangGraph, CrewAI · ChromaDB, pgvector, FAISS · Whisper, IndicConformer, VITS · FastAPI, Django · AWS, GCP, Azure\. Currently focused on Indian\-language AI — ASR, TTS, RAG, and agentic systems built for Bharat\-scale problems\.

## Location

Bengaluru

## AI stack

claude code

claude

codex

cursor

hugging face

claudecode

## Work experience

Lead Associate \(AI\) at I\-PAC \(Indian Political Action Committee\), Apr 2025 \- Present · 1 yr 5 mos, Hyderabad, Telangana, India, 1\. Architected and deployed a multilingual voice AI survey agent \(English, Hindi, Tamil, Telugu\) using Vapi, ElevenLabs, Sarvam LLM, Whisper, and Google STT — processing 20 lakh\+ live telephony calls with real\-time transcription, structured JSON extraction, and QA guardrails\. 2\. Built Scheme\.ai, a production RAG chatbot over government schemes, news, and social media using ChromaDB, pgvector, multilingual Sentence\-Transformers \(e5\-large, LaBSE, MPNet\), GPT\-4\.1, and Groq Llama\-3 for sub\-second multilingual policy QA\. 3\. Developed a ABSA pipeline using Qwen3\-8B served on vLLM over A100 GPUs — running party\-wise sentiment extraction and LLM\-generated daily summaries over Tamil political news corpora\. 4\. Built a topic\-sentiment analytics pipeline \(BERTopic \+ HDBSCAN \+ UMAP\) on multilingual embeddings to cluster 100K\+ YouTube, Twitter, Instagram, and Facebook comments into personality\- and issue\-level dashboards\. 5\. Authored a leadership proposal for in\-house multilingual ASR/TTS \(Whisper large\-v3, IndicConformer, VITS \+ HiFi\-GAN\) covering architecture, data strategy, DPDP Act and TRAI compliance, and deployment roadmap\.

AI Data Scientist at VerifiedTalent, Jan 2025 \- Apr 2025 · 4 mos, Toronto, Ontario, Canada, 1\. Architected and launched an AI\-powered job recruitment & talent profile creation assistant, integrating OpenAI’s GPT models to enhance recruiter\-talent interactions, automating job recommendations, and improving hiring efficiency\. 2\. Built an AI\-driven profile enrichment system leveraging LLMs and Named Entity Recognition \(NER\) to extract and validate key candidate attributes, ensuring 100% structured and high\-quality data ingestion\. 3\. Refined AI assistant workflows by reducing OpenAI API latency by 40%, using caching mechanisms and optimized prompt engineering for faster query resolutions\. 4\. Embedded LLM\-powered auto\-correction and validation in recruiter and talent profile workflows, achieving 100% data consistency through structured JSON generation and multi\-stage validation\. 5\. Developed an LLM\-driven resume parsing and ranking system, leveraging SBERT\-based semantic similarity and fuzzy matching, improving candidate scoring accuracy for recruiters\. 6\. Optimized prompt engineering strategies, designing task\-specific prompts that improved AI\-generated response accuracy and coherence in talent evaluation\.

Data Engineer at AIML Analytics Solutions Pvt Ltd \(OpenDataFabric\), Sep 2023 \- Mar 2024 · 7 mos, Bengaluru, Karnataka, India, 1\. Spearheaded the integration and deployment of LLM\(Llama2\),an advanced conversational AI model, onto server infrastructure, resulting in improvement in customer interaction efficiency and enhancing user experience\. 2\. Designed,developed,and deployed robust backend systems and APIs using Python and Django on Google Cloud Platform \(GCP\),achieving reduction in processing time and greatly enhancing system performance and reliability\. 3\. Integrated comprehensive loan APIs and secure payment gateways to facilitate seamless, efficient, and secure collection of loan amounts, thereby increasing overall financial operations and reduction in payment processing errors\. 4\. Utilized version control systems such as Git and CI/CD pipelines for efficient code management and deployment automation, leading to reduction in deployment errors and increase in deployment speed\. 5\. Pioneered a new underwriting process by analyzing detailed financial statements and transactional data to assess risk and profitability, leading to increase in loan approval accuracy and optimizing the financial decision\-making process\.

Machine Learning Intern at SkillVertex, Feb 2022 \- Mar 2022 · 2 mos, 1\. Learned the basics of high\-level programming languages using a simple introduction into Python programming\. 2\. Understood the various modules like NumPy, Pandas, etc used to build ML systems\. 3\. Got an introduction into the world of ML and learnt about various tools\. 4\. Learnt to handle datasets with data preprocessing\. 5\. Got a deep knowledge about data analytics and evaluation\. 6\. Worked on 2 live projects from ArtifIntel\.

## Education

Master of Technology \- MTech, Metallurgical Engineering, Indian Institute of Technology, Madras

Nanodegree, Artificial Intelligence, Udacity

## Public links

- github: https://github\.com/vinaymh2025

[View the original profile](https://tal.club/meet/vinay-m-h)

```json
{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Person",
      "@id": "https://tal.club/meet/vinay-m-h#person",
      "name": "Vinay M H",
      "url": "https://tal.club/meet/vinay-m-h",
      "image": "https://tal.club/meet/vinay-m-h/profile-photo.jpg",
      "description": "Meet Vinay M H, Multilingual Voice AI & LLM Engineer | RAG · Agentic AI · LLMOps | IIT Madras based in Bengaluru. AI stack includes claude code, claude, and codex. Explore public work links including github.",
      "jobTitle": "Lead Associate (AI)",
      "worksFor": {
        "@type": "Organization",
        "name": "I-PAC (Indian Political Action Committee)"
      },
      "hasOccupation": {
        "@type": "Occupation",
        "name": "Lead Associate (AI)"
      },
      "homeLocation": {
        "@type": "Place",
        "name": "Bengaluru"
      },
      "alumniOf": [
        {
          "@type": "EducationalOrganization",
          "name": "Indian Institute of Technology, Madras"
        },
        {
          "@type": "EducationalOrganization",
          "name": "Udacity"
        }
      ],
      "knowsAbout": [
        "claude code",
        "claude",
        "codex",
        "cursor",
        "hugging face",
        "claudecode"
      ],
      "sameAs": [
        "https://github.com/vinaymh2025"
      ],
      "mainEntityOfPage": {
        "@id": "https://tal.club/meet/vinay-m-h#webpage"
      }
    },
    {
      "@type": "ProfilePage",
      "@id": "https://tal.club/meet/vinay-m-h#webpage",
      "url": "https://tal.club/meet/vinay-m-h",
      "name": "Vinay M H on Tal",
      "description": "Meet Vinay M H, Multilingual Voice AI & LLM Engineer | RAG · Agentic AI · LLMOps | IIT Madras based in Bengaluru. AI stack includes claude code, claude, and codex. Explore public work links including github.",
      "dateModified": "2026-09-09T04:55:18.516Z",
      "isPartOf": {
        "@id": "https://tal.club/#website"
      },
      "mainEntity": {
        "@id": "https://tal.club/meet/vinay-m-h#person"
      },
      "primaryImageOfPage": {
        "@type": "ImageObject",
        "url": "https://tal.club/meet/vinay-m-h/profile-photo.jpg",
        "contentUrl": "https://tal.club/meet/vinay-m-h/profile-photo.jpg",
        "caption": "Vinay M H profile photo on Tal"
      },
      "inLanguage": "en-IN"
    }
  ]
}
```