Arnav Sharma Email: **********.***@*****.***
LinkedIn: arnav-sharma-14b942211 Mobile: +91-701******* Education
Kalinga Institute of Industrial technology Bhubaneshwar, India Bachelor of Technology - Computer Science and Engineering; GPA: 8.39 June 2020 - Sep 2024 Skills
Languages: Python, SQL, JAVA
Frameworks: TensorFlow, Keras, PyTorch, Scikit-learn, NLTK, SpaCy, LangChain, Langgraph, OpenCV, Pandas, NumPy, Matplotlib, Seaborn,Transformers
Core Competencies: Machine Learning, Deep Learning, Natural Language Processing (NLP), Computer Vision, Generative AI, Agentic AI, Retrieval-Augmented Generation (RAG), Large Language Models (LLMs), ANN, CNN, RNN Soft Skills: Leadership, Public Speaking, Event Management, Technical Writing, Time Management Experience
Coforge Greater Noida, India
Software Developer Jul’24 – Present
LLM-Powered Intelligent Document Processing System Tools & Technologies: Python, NLP, Azure Document Intelligence, GPT-4.1, DistilRoBERTa
Designed and developed an end-to-end document processing pipeline for document classification and structured information extraction.
Fine-tuned DistilRoBERTa for document classification, achieving the highest accuracy among multiple evaluated models.
Integrated Azure Document Intelligence OCR to convert unstructured documents into machine-readable text.
Developed a GPT-4.1 prompt-based extraction pipeline to extract structured information from complex business documents.
Implemented rule-based post-processing and validation logic to improve extraction accuracy and consistency.
Improved overall pipeline performance by combining ML-based classification, LLM-driven extraction, and rule-based validation.
Custom Logic-Based Intelligent Document Processing System Tools & Technologies: Python, NLP, Azure Document Intelligence, Regex, DistilRoBERTa
Developed an end-to-end Intelligent Document Processing (IDP) pipeline for document classification and structured data extraction.
Fine-tuned DistilRoBERTa for accurate document type classification and integrated Azure Document Intelligence OCR to extract text from scanned and unstructured documents.
Built a rule-based extraction engine using keyword-based logic and Regular Expressions (Regex) on OCR-extracted text, along with validation rules to improve extraction accuracy and reliability. Hbl Power Systems Hyderabad, India
Data Scientist jun’23 – jun’24
Collaborated with service engineers to analyze operational issues and identify root causes using data-driven approaches.
Assisted in implementing process improvements that enhanced operational efficiency and reduced issue resolution time.
Projects
Equity News Decoder
Python, Streamlit, LangChain, OpenAI API, FAISS
Developed an LLM-powered news research application that enables users to analyze multiple news articles and perform natural language question answering.
Built an end-to-end RAG pipeline using LangChain, OpenAI Embeddings, and FAISS for semantic search and context-aware information retrieval.
Automated URL-based article ingestion, text extraction, recursive text chunking, embedding generation, and FAISS vector indexing for efficient knowledge retrieval.
Designed an interactive Streamlit interface supporting real-time analysis of multiple news URLs with source-backed responses to improve answer reliability and reduce hallucinations. Awards and Certifications
• Impact Award and got a rating for timely delivery of work- Sep, 2025
• Hackerank: Intermmediate level(Problemsolving), Basic level(ProblemSolving)