Jyotidip Barman

Spearheaded the design and development of a real-time marine litter detection system using advanced Convolutional Neural Networks (CNNs) integrated with satellite imagery.

Engineered a patch-based CNN architecture optimized for segmentation and classification of marine debris across diverse oceanographic and environmental conditions.

Collaborated with marine scientists and environmental data experts to validate AI predictions against ground-truth data, improving the ecological reliability of the model.

Implemented data augmentation and ensemble methods to enhance generalization across regions and seasons.

Tuned hyperparameters using cross-validation techniques, leading to substantial gains in precision (increased by 12%) and recall (increased by 17%).

Contributed to policy-driven marine health assessment by providing AI-generated spatial maps to coastal monitoring agencies.

Designed and implemented a Retrieval-Augmented Generation (RAG) pipeline utilizing OpenAI GPT-4, LangChain, and FAISS vector database for enterprise knowledge search.

Integrated semantic search and embedding-based retrieval to enhance accuracy in contextual enterprise Q&A.

Developed internal copilots leveraging company knowledge bases to assist sales and customer service teams.

Designed backend services for LLM orchestration pipelines, including prompt chaining, document ingestion, and embedding storage.

Developed scalable backend microservices using Python FastAPI and Node.js with MongoDB integration.

Deployed model inference pipelines on Azure (AKS, ACI) and managed cloud infrastructure with Docker and Jenkins.

Designed scalable data architecture utilizing Azure ADLS and Databricks.

Automated ETL pipelines across Bronze/Silver/Gold zones using Databricks Workflows.

Integrated secure access via Unity Catalog and delivered executive dashboards with Power BI.

Developed a novel hybrid deep learning architecture combining Residual Neural Networks (ResNet) with a patch-based CNN approach for detecting pneumonia from chest X-ray scans.

Created a preprocessing pipeline including NLM denoising, CLAHE enhancement, and discrete wavelet transform (DWT) to standardize radiographic inputs.

Incorporated a Large Language Model (LLM) to interpret model outputs and generate explainable summaries to assist radiologists in diagnosis.

Achieved a training accuracy of 99% and 93% test accuracy on an independent dataset with robust cross-validation (5-fold) and undersampling strategies to handle class imbalance.

Benchmarked performance against 13 state-of-the-art models and published results in internal medical imaging journals.

Migrated SAS platform to Azure with Knime and Power BI.

Built data models from scratch, integrated marts, and enforced data governance.

Designed and implemented Natural Language Processing (NLP) pipelines to extract customer sentiment from large-scale feedback datasets including surveys, chat logs, and social media.

Trained transformer-based models (BERT, RoBERTa) and LSTM classifiers to identify emotion, tone, and urgency in multilingual feedback streams.

Integrated Named Entity Recognition (NER), topic modeling (LDA), and keyword analysis to generate actionable insights for product and CX teams.

Delivered a dashboard-driven solution enabling real-time tracking of Net Promoter Score (NPS) shifts and sentiment trends across customer segments.

Generated insights that directly contributed to a 15% increase in customer satisfaction (CSAT) and a 20% reduction in support churn through targeted improvements.

Led SQL, QlikView, and SSRS-based analytics initiatives for key clients including Sephora, Diageo, and Wells Fargo.

Created T-SQL modules, planogram checklists, and dashboards for comprehensive enterprise reporting.

Migrated legacy BI systems to modern technology stacks across diverse client portfolios.

Facilitated seamless customer integration into the Salesforce platform, enhancing onboarding and adoption processes.

Developed custom automations to maximize Salesforce feature utilization, significantly boosting efficiency and productivity.

Provided comprehensive training on Salesforce Development, Administration, Data Analytics tools, Database Management, and Visualization tools to team members.

Created impactful reports and visualizations using QlikView, Power BI, and SSRS, delivering valuable insights to stakeholders.

Supported the development and maintenance of Power BI and Power BI Reporting Server, ensuring robust reporting capabilities.

Provided development and support for SAS Enterprise Guide, SAS Enterprise Miner, and Aperture Data Studio, empowering data analysis and decision-making.

Replaced Excel-based solutions in the JDA space with automated T-SQL modules, improving data processing and accuracy.

Migrated SSIS and SSAS components from SQL Server 2008 to SQL Server 2016, ensuring compatibility and leveraging new features.

Conducted a proof-of-concept (POC) pilot project for Big Data and Hadoop, exploring potential for data handling and integration with Talend and MapR technology.

Implemented data mining projects using R Studio, including sentiment analysis of customer feedback, language translation, and speech analysis for WordClouds.

Developed and enhanced MicroStrategy reports, enabling advanced analytics and data visualization capabilities.

Created SQL Server T-SQL and JDA modules to support database operations and optimize performance.

Supported the development and enhancement of Actuate reports, contributing to a pilot project for Actuate migration using SSRS, OBIEE, and MSTR.

Optimized Actuate reports and Oracle PL/SQL queries for improved efficiency and performance.

Contributed innovative ideas to the team and played a hands-on role in handling ad hoc database queries and resolving production issues.