Rakshith Ravisankar

TDP Associate @ Optum

💻 Technical Skills

🧠 Programming Languages

  • C++ – Strong understanding of DSA concepts and problem-solving
  • Python – Data analysis, ML model building, research projects
  • Java – Used in backend development with Spring Boot (internship)

🛢️ Databases

  • Microsoft SQL Server, MySQL, Oracle SQL – Experienced in writing queries and database management

📦 Packages & Libraries

  • pandas, NumPy – Data preprocessing, analysis
  • matplotlib, seaborn – Visualizations for reports & model evaluation
  • Scikit-learn – Classical ML algorithms and preprocessing

📊 Visualization Tools

  • Power BI – Dashboards and visual summaries

⚙️ Frameworks

  • TensorFlow, Keras – Deep learning model development (CNNs, RNNs)
  • Spring Boot – Built a CRUD-based internal system during internship

🌐 Web Technologies

  • HTML, CSS – Basic frontend customization and styling

🔁 Version Control

  • Git, GitHub – Project versioning, collaboration, and hosting

🧪 Projects

Thermal Imaging-based Human Activity Recognition Using YOLOv11

Designed a human activity recognition system using YOLOv8 and YOLOv11 for thermal images to classify into 9 activity classes

  • Achieved high bounding box accuracy with YOLOv11l (mAP@50-95 = 0.90276) and strong pose detection accuracy with YOLOv11s (mAP@50-95 = 0.82407), with most predictions above 90% confidence
  • YOLOv8s further outperformed v11, achieving 0.91506 (box) and 0.84857 (pose), showing its superior performance for thermal-based HAR

Multi-Drone Classification Using 5G RSSI Data

Built ML and DL models to classify the number of drones using wireless RSSI signals segmented into time windows. Evaluated various algorithms for real-time drone detection.

  • Achieved 89.72% accuracy using XGBoost and 89.38% with CNN
  • Demonstrated that simpler models can outperform deeper networks for signal-based classification tasks

ARTA: Attention-guided Reinforcement Trading Agent

Developed a deep learning-based trading system using BiLSTM layers and attention mechanisms for stock price forecasting. Integrated model predictions into a reinforcement learning environment to simulate and optimize trading decisions.

  • Achieved an average R² of 0.8846 and peak R² of 0.9793 across 10 Nifty 50 stocks
  • Outperformed baseline models like LSTM, Transformer, and hybrid models like CNN-LSTM-GRU

🏅 Certifications

  • Learning Analytics Tools (NPTEL) – IIT Bombay
    Achieved Elite Silver (Ranked 32 out of 835) in a 12-week course on educational data analytics and learning metrics

  • Introduction to Industry 4.0 and Industrial Internet of Things (NPTEL) – IIT Kharagpur
    Achieved Elite in a 12-week course covering smart manufacturing, IIoT technologies, and automation

  • Ethical Hacking (NPTEL) – IIT Kharagpur
    Achieved Elite Silver in a 12-week course focused on system vulnerabilities, attack types, and ethical security practices

  • Privacy and Security in Online Social Media (NPTEL) – IIIT Hyderabad
    Achieved Elite Silver in a 12-week course exploring digital privacy

  • Introduction to Large Language Models (LLMs) (NPTEL) – IIT Delhi
    Achieved Elite in a 12-week course introducing transformer architectures and foundational concepts behind LLMs