Shubham kedia

• Designed and deployed generative AI solutions leveraging GPT and LangChain to transform client operations across healthcare and financial services.
• Developed an OCR and NLP pipeline using Azure AI services to process and analyze unstructured document data from scanned PDFs and images.
• Created a fully automated SAP onboarding presentation system using a multi-agent framework, reducing client onboarding time by 80%.
• Deployed AI models in production environments using Docker and Kubernetes, ensuring scalability and reliability.
• Led cross-functional teams to deliver impactful AI-driven solutions and mentored junior data scientists.

Achievements:

• Implemented a data pipeline integrating OCR outputs with NLP models, improving text summarization accuracy by 25%.
• Improved project delivery speed by 50% through insight-driven project management.

• Built scalable ETL pipelines and credit risk models using Apache Spark and Python, enhancing loan approval decisions.
• Deployed NLP models for sentiment analysis and document understanding, reducing reliance on bureau data.
• Conducted ARIMA-based time series forecasting for financial planning, boosting operational efficiency.

Achievements:

• Reduced credit defaults by 7% with a predictive risk assessment model.
• Increased Gross Transaction Value by 5% through targeted customer engagement campaigns.

• Led the development of AI models for IoT-enabled smart appliances, focusing on Non-Intrusive Load Monitoring (NILM) for energy consumption forecasting.
• Conducted customer segmentation and RFM modeling, resulting in a 15% increase in customer retention.

Achievements:

• Boosted energy consumption forecasting accuracy by 60% through AI models.
• Migrated on-premise infrastructure to cloud-based solutions, reducing costs by 20%.

• Developed advanced credit risk models for banks across Southeast Asia, leveraging alternative data sources.
• Created self-service analytics tools for real-time insights into financial performance.
• Led Python training sessions for 50+ professionals across Asia-Pacific.

Achievements:

• Enhanced model accuracy by 15% through robust data validation processes.
• Improved financial predictions and resource allocation accuracy for a Japanese payment processing company.