🎓  I recently graduated with a Master's in Data Science from the University of Michigan (GPA: 3.53/4.0), building on my Bachelor of Technology in Computer Science and Engineering from PES University (GPA: 3.53/4.0).

🚀  My passion lies in leveraging data to build intelligent systems and solve complex problems. I have a strong foundation in Machine Learning, Large Language Models, Information Retrieval, and Data Engineering.

💡  I've gained valuable hands-on experience through impactful roles, including:

• Software Development Engineer Intern @ Amazon: Engineered a distributed locking mechanism using Amazon DynamoDB for AWS Batch Scheduler, significantly improving system reliability and reducing troubleshooting time.
• Data Scientist @ Genpact: Developed customer segmentation models that led to a 30% lower churn and 25% higher average order value.
• Graduate Student Instructor @ University of Michigan: Led discussions and created lab assessments for "Introduction to Statistics and Data Analysis," enhancing student comprehension by 20%.

🌱  I'm continuously exploring advances in Transformers, Self-Attention Mechanisms, and scalable MLOps practices.

✍️  In my free time, I enjoy playing football, table tennis, and reading philosophy books.

✉️  Feel free to drop me an email at aryanrajeshsharma@gmail.com with a subject containing "GitHub: ..."! I'd be delighted to connect.

• Machine Learning & AI: TensorFlow, Keras, PyTorch, Hugging Face, Scikit-learn, NumPy, Pandas, LLMs, Deep Learning, NLP
• Data Engineering & Big Data: Apache Spark, Apache Airflow, AWS (DynamoDB, S3, Batch, CloudWatch, QuickSight), Docker, Kubernetes, Hadoop, Kafka
• Programming Languages: Python, R, SQL, Java, C/C++, JavaScript, C#, Go, TypeScript
• Databases: MySQL, MongoDB, CosmosDB, Cassandra, PyMongo
• Cloud Platforms: AWS, GCP, Azure
• Data Visualization: Tableau, Power BI, Matplotlib, Seaborn
• Tools & Others: Git, Linux/Unix, JIRA, Agile Methodologies

• GDINOSAUR - Grounding DINO with Spatial Awareness for REC
  • Co-engineered a data pipeline for RefCOCO-3DS, generating a 7,000+ image synthetic dataset with Blender.
  • Advanced Grounding DINO's Referring Expression Comprehension by fine-tuning, achieving a 15% mAP uplift in spatial/non-spatial understanding.
  • (EECS 545 Course Project, University of Michigan)
• Real-Time Two-Handed Sign Language Translation (RtTSLC)
  • Pioneered a deep learning framework (CNN, Siamese Networks) achieving 98% accuracy and <100ms latency for real-time, two-handed Indian Sign Language translation.
  • Published in Springer & presented at an International Conference. (See Publications below)
• ECG Anomaly Detection for Wearable Devices
  • Architected and validated a deep-learning model (TensorFlow-Keras) for detecting cardiac anomalies from 2-lead ECG signals, achieving 98% predictive accuracy.
  • Presented at the 13th HiPC Student Research Symposium (SRS).

• RtTSLC: A Framework for Real-Time Two-Handed Sign Language Translation
  • Springer, June 2023. DOI: 10.1007/978-981-99-0769-4_62
  • Pioneered a deep learning framework (CNN, Siamese Networks) achieving 98% accuracy for real-time Indian Sign Language translation.
• Sign Language Translation Systems: A Systematic Literature Review
  • IGI Global, October 2022. DOI: 10.4018/IJDCF.311448
  • Synthesized insights from 200+ papers, identifying key research gaps and defining novel methodological avenues.
• Cardiac anomaly detection models for wearable devices
  • 13th HiPC Student Research Symposium (SRS), part of 28th IEEE International Conference on High-Performance Computing, Data, & Analytics, October 2021.
  • Architected and validated a deep-learning architecture for detecting cardiac anomalies, demonstrating 98% predictive accuracy.