Featured Projects

Here are some of the key AI and Machine Learning projects I have built.

• Operationalizing Fetal Ultrasound Analysis
  An End-to-End ONNX-PyTorch Pipeline for Abdominal Structure Segmentation and Risk Stratification.. Project Demo
• SafePredict : Building a Real-Time “Shadow Monitor” for Humanoid Robot Safety
  How to bridge the gap between open-source robotics data (LeRobot v0.6.0) and enterprise MLOps (Triton/Kafka) to prevent physical AI failures. Code
• Traffic Signs Detection with YOLOv8
  Real-time object detection model optimized for edge devices. Code
• Medical Test Analyzer
  AI pipeline for analyzing and interpreting complex medical test data. Code
• Multimodal-Code-Architecture-Co-Pilot-MLOps-Agent
  Multimodal-Code-Architecture-Co-Pilot-MLOps-Agent-MLOps Best Practices Code
• Multi-Tenant-LLM-Serving-Platform-on-Kubernetes
  A self-service platform where multiple “tenants” (teams) can deploy different open-source LLMs behind a unified API, with auto-scaling, rate limiting, and usage tracking Code
• Supply-Chain-Risk-Intelligence
  A production‑ready MLOps pipeline that simulates real‑time IoT sensor data, detects anomalies with Spark + Kafka, manages models with MLflow, orchestrates retraining with Kubeflow, and monitors everything – built for AI‑driven smart city and industrial IoT solutions. Code
• ASNR-on-lumbar-spine-degenerative-conditions
  Analysis of a Deep Learning Framework for Automated Diagnosis of Lumbar Spine Degenerative Conditions from MRI Data Code
• Analysis of Deep Learning Approaches for Atmospheric Physics Modeling Using TFRecords
  The provided Jupyter notebook (atmospheric-physics-with-tfrecords.ipynb) demonstrates a comprehensive deep learning workflow for climate and atmospheric physics modeling using TensorFlow and TFRecord datasets. The implementation showcases modern machine learning practices applied to climate science, specifically addressing the LEAP Atmospheric Physics AI ClimSim challenge dataset containing 556 input features and 368 target variables. Code
• Smart Order Dispatcher
  Reinforcement learning system for optimizing logistics and delivery routing.
• NLP Search Engine & LLM Q&A
  RAG-based search engine utilizing Large Language Models for accurate Q&A. Code
• Lip-Syncing Audio to Video
  Computer vision and audio processing pipeline for precise video dubbing.