Headline

Data Scientist | NLP | Machine Learning | IIIT B| LJMU

About

Results-driven Data Scientist with 5+ years of experience in optimizing business operations through cutting-edge NLP techniques. I help companies revolutionize their processes and achieve remarkable outcomes. • Implemented advanced technologies including LLMs, GenAI, and NLP, utilizing data formats with Qdrant VectorDB and integrated LLM models such as GPT-3.5, GPT-4, Mistral, Claude, and Llama. • Implemented RAG model with BGE-Large-Embeddings, establishing robust pipelines for RAG and LLM evaluations to Improve Retrieval results by applying Query Rewriting, meta filtering, Reranking. • Trained Legal BERT Classifier for extracting Legal clause from 250+ classes and designed an extraction pipeline with 90% Accuracy. • Reduced the MLOPS overhead by 20 hrs per week by implementing data drift models, model evaluations and data standardization • Spearheaded implementation of text classification and table standardization frameworks, reducing manual invoice processing time by 50% and increasing accuracy by 40%. • Trained NER models for bank statements and income computation, achieving 97% accuracy in extracting salary data. • Enhanced document and image classification models, improving accuracy by 5% and reducing manual effort by 50 hours. Let's collaborate to drive transformative solutions and propel your business forward! Industry Experience : • Manufacturing • Oil & Gas • Insurance Tech • Legal Tech --- Technical Skills Frameworks: Transformers (Hugging Face), scikit-learn, Keras, TensorFlow, PyTorch, HF Optimum, FastAPI, Flask NLP: NER, Sentiment Analysis, Text Classification, Flair, NLTK, SpaCy, Gensim, CRF++, HF Transformers, Langchain Deep Learning: LSTM, GPTs, BERT, RoBERTa, T5, DiT, Prompt Engineering with GPTs, LLMs, RAG, RHLF, LoRA Cloud Computing: Microsoft Azure, AWS Programming Languages: C, C++, Python, SQL Web Scraping: Beautiful Soup (bs4), Scrapy

Location

Maharashtra

AI stack

claude code

codex

hugging face

Work experience

Senior Machine Learning Engineer at IDfy, Mar 2025 - Present · 1 yr 7 mos, On-site

Data Scientist at ContractPodAi, Jul 2023 - Apr 2025 · 1 yr 10 mos, Mumbai, Maharashtra, India, - Integrated cutting-edge LLMs (GPT-3.5, GPT-4,Claude, Mistral, LLaMA) with Qdrant VectorDB to enhance semantic search and retrieval. - Built robust Retrieval-Augmented Generation (RAG) pipelines leveraging OpenAI embeddings, improving retrieval accuracy through advanced query rewriting, meta-filtering, and reranking strategies. - Trained a Legal-BERT classifier to extract legal clauses across 250+ classes, achieving 90% accuracy. - Actively led requirement discussions with clients and played a key role in mentoring and guiding junior team members.

Data Scientist at i3systems.ai, Mar 2022 - May 2023 · 1 yr 3 mos, Mumbai Metropolitan Region, Motor Invoice Digitization •Led the implementation of Azure OCR and Form Recognizer to automate motor invoice processing, resulting in a 50% reduction in manual effort and a 40% increase in accuracy. • Trained a Flair model to accurately extract meta entities from invoices, such as vendor names, invoice dates, GST numbers, and chassis numbers, Financial Underwriting • Developed and trained NER models using multiple libraries and frameworks, including flairnlp, NLTK, and CRF++, to extract relevant information from Bank Statements, Form16, and Computation of Income (COI). • Deployed an algorithm that extracted salary data from bank statements with 97% accuracy. Medical Examination Reports • Trained a regression model to improve data extraction efficiency by 15% by finding data point coordinates. • Trained a Resnet-18 model to accurately classify checkboxes, signatures, and stamps, further streamlining the medical examination report process. Enhancement • Improved the accuracy score of document and image classification models by 5% through an upgrade. • Customized an auto-learning framework for profile terms with range values, reducing compute time by 7 seconds per case and decreasing manual user intervention by 50 hours.

Senior Software Engineer at LTI - Larsen & Toubro Infotech, Nov 2020 - Mar 2022 · 1 yr 5 mos

Associate Consultant at LTI - Larsen & Toubro Infotech, Aug 2020 - Oct 2020 · 3 mos, Mumbai, Maharashtra, India

Graduate Engineering Trainee at LTI - Larsen & Toubro Infotech, Aug 2019 - Jul 2020 · 1 yr, Airoli

Data Science Intern at Agrima Infotech PVT LTD, Oct 2018 - Jan 2019 · 4 mos, Remote, - Developed a solution for resolving issues with a unit converter feature in an app using advanced text preprocessing and cleaning techniques like tokenization and stemming. - Implemented Spacy named entity recognition (NER) to accurately identify and convert units, and employed a spellcheck model to ensure the accuracy of 85% - Hosted the solution on a server to ensure easy accessibility and provide a seamless user experience.

Education

Master's degree, NLP, Liverpool John Moores University

Postgraduate Degree, Data Science, International Institute of Information Technology Bangalore