Job highlights

- 3+ years experience with Python, TensorFlow, PyTorch, LangChain, and expertise in Generative AI, LLMs, RAG, and prompt engineering
- Develop and deploy GenAI applications including document QA chatbots, RAG-based summarization, semantic retrieval pipelines, and cloud-native scalable LLM services

Technical Skills: Programming Languages: Python, SQL

Frameworks: Tensorflow, PyTorch, Keras, Langchain, Streamlit, LangGraph

Data Science and Miscellaneous Technologies: Deep Learning, Natural Language Processing, Machine Learning, Artificial Intelligence, Generative AI, LLMs, RAG, Transformers, Finetuning LLM, Prompt Engineering, Vector Database, Computer Vision, CNN, Regression, LSTM, Random Forest, Data Analytics, Optimisation techniques, Data Structures, Algorithms

Tools: Numpy, Pandas, Matplotlib, Scikit-learn, HuggingFace, Docker, ETL, NLTK, Spacy, Linux, Git

Qualifications: Bachelors in Technology / Masters in Data Science/ Masters in Mathematics/ Masters in Physics/ Statistics or any other quantitative field

Relevant Work Experience: 3+ years

Projects:

1. Generative AI

a. Developed an interactive StreamLit interface for Document Question Answering Chatbot for large PDFs using optimised chunking strategies, LLaMa2 model, FAISS DB, LangChain b.



Built a Semi-Structured RAG based Chatbot for complex Tabular Data Summarization and QA by integrating image preprocessing and OCR; improved accuracy by custom prompt engineering with reduced inference time of 10seconds. c. Improved LLM response quality by reducing hallucinations, enhanced accuracy using apt evaluation metrics.
1. Built and deployed a cloud-native GenAI E-commerce Assistant using RAG, MCP, and LangChain agents to combine static catalog data with real-time marketplace APIs for intelligent product search and comparison.

a. Developed scalable semantic retrieval pipelines (FastAPI + vector DB) delivering context-aware recommendations with low latency and reducing manual product research time by ~75%. b. Implemented end-to-end CI/CD and secure AWS deployment (Docker, EKS, GitHub Actions) enabling auto-scalable, production-ready LLM services.
1. Implemented a self-supervised learning pipeline for defect classification using Python and PyTorch, eliminating the need for extensive labeled datasets in manufacturing domain. Achieved 81.5% classification accuracy over 20K augmented images significantly reducing manual data annotation efforts

📌 Gen AI Data Scientist (Gurugram)
🏢 hrindiasolutions.co.in
📍 Gurugram