Open to opportunities

shambhavi kumari

@shambhavikumari

Message

AI/ML builder specializing in RAG, NLP, and production-ready LLM applications for real-world problem solving.

India

Message

What I'm looking for

I’m looking for an entry role where I can build production-grade RAG and NLP systems—deploy LLM workflows with tools like FastAPI/Vertex AI, collaborate closely on grounded retrieval, and continuously improve accuracy, robustness, and usability.

I’m building AI systems that move beyond demos—especially Retrieval-Augmented Generation (RAG) and NLP pipelines that can be grounded in real documents. In NYAAI, I engineered a full-stack AI legal justice platform using RAG, LangChain, and LLM APIs (Groq + Gemini) to detect legal violations and auto-generate court-ready notices, reducing manual drafting effort by ~80%.

I also focus on production-ready delivery and measurable performance: LPU Knowledge Assistant delivered context-aware Q&A across 13 LPU policy documents with hallucination mitigation, and Serenova fine-tuned BERT to classify text across 5 psychological distress levels with 83%+ accuracy—deployed with a live Gradio demo on HuggingFace Spaces. I’m continuously expanding into deployment and cloud workflows through Generative AI learning paths and Vertex AI work with real application integration.

Experience

Work history, roles, and key accomplishments

Generative AI Essentials

IBM SkillsBuild

May 2026 - Present (2 months)

Completed Generative AI Essentials focused on using LLMs to work with data.

Generative AI and LLMs

Google Cloud Skills Boost

May 2026 - Present (2 months)

Completed Google Cloud’s Generative AI learning path covering LLM architecture, transformer models, diffusion models, and prompt engineering. Applied RAG and vector-based search, and deployed GenAI models using Vertex AI for application workflows.

LLM Architecture Transformer Models Diffusion Models Prompt Engineering Retrieval Augmented Generation (RAG)Vector Search Vertex AI Google Cloud LLM

Claude API Building

Anthropic

May 2026 - Present (2 months)

Completed the course focused on building with the Claude API and core generative AI development concepts using LLMs.

NYAAI AI Legal Justice

GitHub

May 2026 - Present (2 months)

Engineered a full-stack AI legal justice platform using RAG, LangChain, and LLM APIs (Groq + Gemini) to detect legal violations and auto-generate court-ready notices. Architected a grounded RAG pipeline with vector database retrieval and integrated IndicNLP and OCR for multilingual scanned legal document processing.

Full Stack Retrieval Augmented Generation (RAG)Lang Chain GROQ Google Gemini Vector Databases PostgreSQL Celery REST APIs IndicNLP Grounded Information Extraction

Microsoft Azure AI Fundamentals

Microsoft

Apr 2026 - Present (3 months)

Completed Microsoft Azure AI Fundamentals (AI-900) covering foundational AI concepts and Azure AI capabilities.

Intermediate Machine Learning

Kaggle

Apr 2026 - Present (3 months)

Completed an Intermediate Machine Learning course on Kaggle.

LPU Knowledge Assistant

GitHub

Mar 2026 - Present (4 months)

Architected an end-to-end RAG-based QA system using LLaMA 3.1 8B via Groq API and ChromaDB for context-aware answers across 13 LPU policy documents. Built a semantic search pipeline with sentence-transformers, and developed a production-ready FastAPI backend with .docx ingestion automation to reduce hallucinations.

Retrieval Augmented Generation (RAG)GROQ ChromaDB Sentence Transformers Semantic Search fastAPI REST APIs NLP

Serenova AI Distress Detection

GitHub

Feb 2026 - Present (5 months)

Fine-tuned a BERT (110M) model in PyTorch using HuggingFace Transformers to classify text into 5 psychological distress levels, achieving 83%+ accuracy. Deployed the fine-tuned model on HuggingFace Spaces with a live Gradio demo for real-time distress detection.