---
title: "vishal sharma - Lead AI Engineer Products & Platform in Bengaluru | Tal"
description: "Meet vishal sharma, Lead AI Engineer | 5+ years of experience | Ex Samsung R&D | Knowledge Graphs | Agentic AI | RAG | Multi-Agent Systems | Inference Optimization based in Bengaluru. AI stack includes langgraph, langsmith, and openai."
image: "https://tal.club/meet/vishal-sharma-11/share-image"
---

# vishal sharma on Tal

## Headline

Lead AI Engineer \| 5\+ years of experience \| Ex Samsung R&D \| Knowledge Graphs \| Agentic AI \| RAG \| Multi\-Agent Systems \| Inference Optimization

## About

I am an Applied AI Engineer and Data Scientist with 5\+ years of experience building Machine Learning, Generative AI, and Retrieval\-Augmented Generation \(RAG\) solutions that drive measurable business impact\. My expertise spans Agentic AI systems, large\-scale information retrieval, recommendation systems, experimentation platforms, and demand forecasting, with a strong focus on taking AI solutions from research to production\. Currently at Samsung R&D, I lead the development of AI\-powered customer support automation systems\. I designed and deployed an end\-to\-end Agentic RAG platform combining intent classification, hybrid retrieval, cross\-encoder reranking, answer generation, and LLM\-based evaluation, reducing Mean Time to Resolution \(MTTR\) by 42%\. I also built and scaled a Llama 70B inference platform on AWS EKS using Kubernetes and vLLM, supporting thousands of customer interactions daily\. In addition, I have worked on recommendation system experimentation frameworks and AI product analytics, helping evaluate feature impact across large user populations\. Prior to Samsung, I worked at Asper AI \(a Fractal company\), where I developed a financial knowledge retrieval platform over 10K\+ annual reports using knowledge graphs, semantic search, and reranking techniques\. The solution improved retrieval performance significantly while supporting large\-scale concurrent workloads\. I also built MLOps workflows with MLflow, CI/CD automation, and reproducible deployment pipelines to accelerate model development and delivery\. My broader experience includes demand forecasting, statistical experimentation, explainable AI, feature engineering, predictive modeling, and cloud\-native ML systems\. I enjoy solving complex business problems using data and AI while ensuring solutions remain scalable, reliable, and production\-ready\. Core Skills: Generative AI, Agentic AI, RAG, LLMs, LangChain, LangGraph, Vector Search, Knowledge Graphs, Machine Learning, Statistics, A/B Testing, XGBoost, Forecasting, PyTorch, Hugging Face, AWS, Azure ML, Kubernetes, Docker, Terraform, MLflow, MLOps, CI/CD\.

## Location

Bengaluru

## AI stack

langgraph

langsmith

openai

huggingface

Chroma

Qdrant

## Work experience

Lead AI Engineer Products & Platform at Incedo Inc\., Oct 2026 \- Present · 1 mo, Bengaluru, Karnataka, India

Lead Engineer at Samsung R&D Institute India \- Bangalore, Sep 2025 \- Aug 2026 · 1 yr, On\-site, Developed an end\-to\-end Agentic RAG pipeline combining intent classification, hybrid retrieval, cross\-encoder reranking, answer generation, and an LLM\-as\-a\-Judge feedback loop that identified missing information and triggered corrective retrieval, reducing MTTR by 42%\. Built and scaled a Llama 70B inference service on AWS EKS using vLLM and Kubernetes\-based orchestration, supporting 3K\+ customer tickets per day and enabling production\-scale deployment of AI\-powered support automation\. Built an end\-to\-end A/B testing framework for Samsung Marketplace recommendation systems, defining experiment hypotheses, user sampling strategy, experiment duration, and statistical success metrics to evaluate CTR improvements\. Performed statistical experimentation and analysis to validate model effectiveness, demonstrating a 12% increase in CTR and enabling rollout of the new recommendation engine to production\. Analyzed post\-launch behavioral data for 500K\+ users to evaluate the impact of the Now Brief AI feature, identifying a statistically significant 14% increase in engagement among active user cohorts using Welch's t\-test and cohort\-based analysis\. Built a feature impact analytics pipeline leveraging feature engineering and XGBoost to identify key engagement drivers, generating actionable insights that informed AI roadmap prioritization and contributed to a 10% increase in feature investment allocation\.

Senior Data Scientist at Asper\.ai, May 2024 \- Aug 2025 · 1 yr 4 mos, Bengaluru, Karnataka, India, Designed, built, and owned a production Retrieval\-Augmented Generation \(RAG\) system deployed on Azure, serving \~100 internal users for financial research use cases\. Implemented semantic vector search, shard\-based retrieval, and scalable reranking pipelines, enabling support for 1,000\+ concurrent queries\. Optimized retrieval performance by introducing batching, parallel shard execution, and ANN optimizations, reducing end\-to\-end latency to \~1 second on GPU\-backed VMs\. Integrated MLflow for experiment tracking and built CI/CD pipelines for model versioning, testing, and deployment\. Improved research productivity by \~70% through faster and more accurate information retrieval\. Worked extensively on vector databases, embedding models, reranking strategies, and evaluation metrics on the retriever side\.

Data Scientist at Asper\.ai, Dec 2022 \- May 2024 · 1 yr 6 mos, Bengaluru, Karnataka, India, • 95% accurate Long Term Forecasting \+1\.5\-2% revenue uplift while transforming category\-level forecasting\. • Developed a SHAP\-based factor contribution method providing clear insights into differences in Year\-over\-Year predictions\. • Developed an Attribute and Proxy based approach while reducing forecast bias by 5% and boosting revenue by $25 million through accurate demand forecasting for New Products\. • Auto\-ML driven forecasting \+5% accuracy \+$16M revenue and a new standard for predictive excellence\. • Developed an Ensembled Time\-Series and XGBoost approach \+10% accuracy \+$20M revenue uplift while transforming brand forecasting\. • Automated pre and post processing data steps slashing refresh time by 66% from 3 days to just 1 day enabling faster insights and decision making\. • Created an innovative product feature that enables clients to input specific feature values facilitating precise factor contribution evaluation and delivering personalized insights that inform strategic choices\.

Data Scientist at Equifax, Oct 2022 \- Dec 2022 · 3 mos, Karnataka, India, • Recreated the XGBoost model on vertex AI while leveraging Google Big Query ML’s scalable infrastructure\. • Conducted a rigorous sample bias analysis to assess performance focusing on capture rates\. • Employed Bayesian Optimization for hyperparameter tuning yielding a \~2% improvement in capture rates\. • Demonstrated expertise in precision model optimization while driving business value through data driven enhancements\.

Data Scientist at Accolite Digital, Oct 2021 \- Sep 2022 · 1 yr, Gurugram, Haryana, India, Developed an XGBoost to simulate weekly marketing spend allocation empowering clients to optimize budget distribution\. • Created a slope\-based optimization approach using Constrained Linear Regression enabling clients to divide the marketing budget across four categories and improve Ad Spend\(ROAS\) by 8\-10%\. • Built a Random\-Forest model with hyperparameter tuning using Optuna to predict cancellation rates for converted leads helping clients to adjust capacity and resources by 15\-20%\. • Developed Log\-Based Linear model while achieving approximately 80% accuracy in predicting jobs while uplifting revenue by 1\.5\-2%\. • Designed a Single Factor Scenario Planning approach allowing clients to simulate marketing spend across channels and forecast job generation impact\.

Data Science Intern at Accolite Digital, May 2021 \- Oct 2021 · 6 mos, India, • Developed a Resume Parser leveraging NLP and Deep Learning methods streamlining resume screening and reducing manual efforts by \~30% • Designed a Context\-Similarity method utilizing Sentence\-BERT transformer embeddings and Siamese Networks to accurately predict the similarity between resumes and job descriptions enhancing candidate matching

## Education

Bachelor's degree, Electrical and Electronics Engineering, kiet group of institutions

[View the original profile](https://tal.club/meet/vishal-sharma-11)

```json
{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Person",
      "@id": "https://tal.club/meet/vishal-sharma-11#person",
      "name": "vishal sharma",
      "url": "https://tal.club/meet/vishal-sharma-11",
      "image": "https://tal.club/meet/vishal-sharma-11/profile-photo.jpg",
      "description": "Meet vishal sharma, Lead AI Engineer | 5+ years of experience | Ex Samsung R&D | Knowledge Graphs | Agentic AI | RAG | Multi-Agent Systems | Inference Optimization based in Bengaluru. AI stack includes langgraph, langsmith, and openai.",
      "jobTitle": "Lead AI Engineer Products & Platform",
      "worksFor": {
        "@type": "Organization",
        "name": "Incedo Inc."
      },
      "hasOccupation": {
        "@type": "Occupation",
        "name": "Lead AI Engineer Products & Platform"
      },
      "homeLocation": {
        "@type": "Place",
        "name": "Bengaluru"
      },
      "alumniOf": [
        {
          "@type": "EducationalOrganization",
          "name": "kiet group of institutions"
        }
      ],
      "knowsAbout": [
        "langgraph",
        "langsmith",
        "openai",
        "huggingface",
        "Chroma",
        "Qdrant"
      ],
      "mainEntityOfPage": {
        "@id": "https://tal.club/meet/vishal-sharma-11#webpage"
      }
    },
    {
      "@type": "ProfilePage",
      "@id": "https://tal.club/meet/vishal-sharma-11#webpage",
      "url": "https://tal.club/meet/vishal-sharma-11",
      "name": "vishal sharma on Tal",
      "description": "Meet vishal sharma, Lead AI Engineer | 5+ years of experience | Ex Samsung R&D | Knowledge Graphs | Agentic AI | RAG | Multi-Agent Systems | Inference Optimization based in Bengaluru. AI stack includes langgraph, langsmith, and openai.",
      "dateModified": "2026-10-06T14:08:35.634Z",
      "isPartOf": {
        "@id": "https://tal.club/#website"
      },
      "mainEntity": {
        "@id": "https://tal.club/meet/vishal-sharma-11#person"
      },
      "primaryImageOfPage": {
        "@type": "ImageObject",
        "url": "https://tal.club/meet/vishal-sharma-11/profile-photo.jpg",
        "contentUrl": "https://tal.club/meet/vishal-sharma-11/profile-photo.jpg",
        "caption": "vishal sharma profile photo on Tal"
      },
      "inLanguage": "en-IN"
    }
  ]
}
```