Sahil Panjwani

Main Tasks include : creating homework assignments, in class quizzes , programming assignments, mentoring students after class and Grading.

Implementing SAR-based ATR using off-the-shelf CNN backbones in PyTorch, applying custom domain-adaptive augmentations (quadratic phase error, speckle simulation, pixel swapping) to bridge the synthetic-to-real gap. Building scalable data pipelines in Python/NumPy for multi-angle, multi-elevation SAR imagery—while architecting a modular framework ready for upcoming CNN fine-tuning, model distillation, and scaling experiments. Executing iterative model evaluation with cross-validation and ablation studies on real SAR samples to drive robustness improvements and inform future model compression and edge-deployment strategies.

• Created and executed automated test scripts using Selenium WebDriver and Java for a client’s website, enhancing testing efficiency. • Identified and documented over 50 bugs, contributing to the early detection and resolution of issues. • Collaborated with the development team to ensure the timely resolution of identified issues, improving overall software quality.

• Developed a LLM based web scraper for image data extraction, enhancing image generation models. • Integrated selenium functionality in the scraper for dynamic JavaScript sites, improving application usability by 15%.

• Automated heart rate data processing with custom Python module, saving 80 hours/month of Manual Work • Enhanced prediction model performance by 20% with new statistical features • Improved Sentiment Analysis ML model accuracy to 83% using RNN and LSTM networks

• Developed a Facial Recognition-based Attendance System achieving 90% accuracy on still images and 85% on video frames. • Collaborated with team members to design and implement computer vision algorithms for image recognition. • Conducted research and testing to optimize the system's performance and accuracy.