• Developed an AI model with 93% accuracy for predicting suspects in Samsung's big data services using defect data, integrated MLOps for continuous training, and reduced manual effort by 35%.
• Implemented a Gen-AI multi-modal system for generating social media captions with hashtags and emojis in Samsung Gallery, using RAG & LLM, and developed an award-winning API for on-device integration.
• Developed a multi-modal AI for real-time age-appropriate content blocking on smartphones, covering audio, video, and text formats, with on-device integration and award recognition.
• Developed an application for on-device TTS, STT, TTT feature verification comparing S24 on-device AI model with Open-Source Models. Generated detailed comparison Excel reports reducing manual effort by 90% and improving accuracy by 20%.
• Developed an application for multilingual relevant text extraction from images via OCR with LLM. Compared results with on-device gallery AI capture feature verification, generating Excel reports. Reduced manual efforts by 70% and enhanced ROI by 50%.
• Implemented video-to-audio conversion and automated S3 bucket uploads, streamlining training of videos in multiple designated categories for MLOps pipeline in Amazon Sage Maker. Generated Excel reports with detailed analysis and corrected data improved model performance by 30% and resulted in a 15% ROI enhancement.

• Fine-tuned and deployed Llama2 7B LLM with TensorRT and CUDA, developed a Streamlit chatbot using LangChain & RAG for advanced document-based user journeys.
• Created a deep learning model for classifying Chest-X-Ray DICOM data, achieving 95% accuracy and optimized inference with FastAPI on Azure ML.
• Developed an anomaly detection system for manufacturing chips using YOLOv5 and Azure Cognitive Services, deploying models with ONNX, PyTorch, and TFLite
• Built a multilabel business document classification model, achieving 95% accuracy, and deployed a Flask API with Azure web app for optimized accessibility.
• Designed and implemented Airflow pipelines for FTP data extraction and complex transformations, integrating with Azure Synapse Analytics and optimizing historical data storage.
• Developed a PySparkML model for credit score bracket prediction with 95% accuracy and deployed it using Flask API.
• Designed Power BI dashboards for deriving insights from credit-related data to enhance bank email marketing.
• Built Azure Data Factory pipelines with ADLS Gen2, Azure SQL Server, and Azure Databricks for efficient ETL and model training, leveraging Apache Spark ML.

• Boosted client fundraising campaign ROI by developing and implementing machine learning models to predict donor behavior and estimate potential donation amounts based on past profiles. Achieved 96% accuracy.

• Leveraged data analysis and data modeling expertise to accurately visualize donor behavior, optimizing fundraising efforts. Provided valuable insights for more targeted and effective campaigns, thereby maximizing ROI.

• Integrated game mechanics to elevate engagement and loyalty in diverse contexts. Leveraged YOLOv5 and Azure Cognitive Services to enhance footballer performance and productivity. Deployed ONNX and TFLite models for mobile apps.

• Developed Power BI dashboards for smartwatch data analysis, driving user engagement and ROI.

• Hands-on Industrial Training on Big Data Technologies: HDFS, Hive, Apache Spark, Apache Kafka, Apache Pig, Python, and Oozie.

• Developed a real-time Twitter sentiment analysis project, updating every 5 minutes with global trending topics or specified keywords. Implemented with Scala and Spark Streaming for efficient data processing.