Raghav Sharma

Data Scientist

West Bengal, India4 yrs experience

Most Likely To Switch

Key Highlights

Expert in developing AI-driven solutions.
Led impactful projects in healthcare automation.
Strong background in NLP and machine learning.

Stackforce AI infers this person is a Data Scientist specializing in AI solutions for the healthcare industry.

Contact

Skills

Core Skills

Machine LearningNatural Language Processing (nlp)

Other Skills

Data AnalysisData ScienceDeep LearningPython (Programming Language)TeamworkTime Management

About

As a Data Scientist at HiLabs, I specialize in developing scalable AI-driven solutions for complex business problems. My expertise spans Machine Learning, Deep Learning, Natural Language Processing, and Computer Vision, with a strong focus on deploying cutting-edge models to drive automation and efficiency. I hold a Dual Degree (B.Tech + M.Tech) from IIT Kharagpur, where I honed my skills in AI research and applied data science across various domains. My experience includes leading high-impact projects, such as building Retrieval-Augmented Generation (RAG) pipelines for US healthcare contracts, fine-tuning transformer-based models, and optimizing large-scale data processing systems. Previously, I have worked with organizations like RudderStack and Moksh.io, where I developed predictive models, automated data workflows, and enhanced decision-making capabilities. My research contributions in Biomedical NLP and sentiment analysis have been recognized at international conferences, and I continue to explore innovative AI applications in diverse fields. I thrive in dynamic environments, collaborating with cross-functional teams to bridge the gap between data and business impact. I am a passionate and self-motivated individual who is always eager to learn and grow. I believe in the power of continuous improvement and take challenges as opportunities for personal and professional development. Let's connect to discuss AI, data science, and transformative technology solutions!

Experience

Hilabs

2 roles

Data Scientist 2

Jul 2025 – Present · 8 mos

Scaled and enhanced the developed RAG pipeline, further reduced manual effort and expanded scope of OCR service across multiple products
▪ Led a cross domain team of 6 to migrate to a scalable solution, integrating a feedback loop to continuously improve automation rates.
▪ Delivered an 80% efficiency gain by creating a self-service solution for attribute extraction configuration and testing in the RAG pipeline.
▪ Built a Unified OCR service for healthcare documents using SmolDocling and Heron models, enabling scalable multi-product use.
▪ Improved OCR model adaptability resulting in 65% decrease in fine-tuning needs by using active learning guided training data selection.
▪ Integrated diverse data sources to extract market pricing insights enabling strategic, data-driven rate negotiations during contracting.

Data ScienceNatural Language Processing (NLP)Machine LearningDeep LearningPython (Programming Language)

Data Scientist 1

Jun 2024 – Jul 2025 · 1 yr 1 mo

Developed a scalable RAG pipeline on US Healthcare contracts in order to extract information to feed into claims processing automation
▪ Implemented a scalable Retrieval-Augmented Generation (RAG) pipeline using Mistral-7B to extract key pricing terms from contracts.
▪ Automated the configuration of 83% of contracts by extracting key contract data and integrating it into the claims handling application.
▪ Deployed a Question Answering chatbot leveraging the RAG pipeline, achieving 90% QA accuracy and enhancing information retrieval.
▪ Reduced SLA for retrieving market level data by 95% by integrating and centralizing 5+ data sources within the chatbot framework.
▪ Secured $1.5M ARR by managing a Fortune 20 client and converting non technical business requirements into technical solutions.
▪ Reduced implementation time by 90% by integrating a self-serve Business Rules Engine to handle more than 1000 business rules.

Data ScienceNatural Language Processing (NLP)Machine LearningDeep LearningPython (Programming Language)

Rudderstack

Machine Learning Engineer Intern

May 2023 – Jul 2023 · 2 mos · Remote

Performed Next Event Prediction using several approaches like LSTMs, Siamese Network and Two Tower Architecture
Worked on the task of lookalike audience prediction to identify the most valuable customers to a firm
Used Universal Sentence Encoder and Next Event Prediction models to capture user understanding for the task

Machine LearningData Analysis

Moksh.io

Data Science Intern

Apr 2022 – Aug 2022 · 4 mos

Worked on the product from the ideation stage and developed it till the MVP stage.
Processed data from multiple Amazon Seller APIs and converted them into a usable format for better understanding and decision making
Performed various operations on the said data to make it useful and insightful for Amazon merchants.
Handled huge amounts of data coming from some of the bigger merchants selling products on Amazon
Getting data ready for use in various Machine Learning models to be developed later for helping merchants

Data ScienceData Analysis

University of ottawa

Research Intern

Jan 2022 – Jun 2022 · 5 mos

Supervisor: Professor Shantanu Dutta, Full Professor (Finance Area) and Telfer Fellow in Global Finance
Objective: Sentiment analysis of news articles related to the Indian Agricultural Industry
Scraped and labeled relevant news articles from 3 renowned news websites for a period of 4 months
Fine-tuned 5 transformer-based language models for sentiment classification into 3 classes. and compared results
The project can be used as a base for further research work and as an aid in decision-making by policymakers

Natural Language Processing (NLP)Data Analysis

Indian institute of technology, roorkee

Research Intern

Feb 2021 – Sep 2021 · 7 mos

Supervisor: Professor Raksha Sharma, Department of Computer Science and Engineering
Objective: Information Extraction in the Biomedical Domain
Worked on the improvement of the BioBERT model in the field of Biomedical Named Entity Recognition
Replaced the ADAMs optimizer in BioBERT with LAMB optimizer in order to significantly reduce the training time
Pre-trained the model from scratch by using the RoBERTa model approach instead of BERT used orgiinally
The project can be used as means of better information extraction in the very fast paced biomedical domain