Objective: Build a multimodal Geo-NLI system for satellite imagery tasks

• Developed TRINETRA to orchestrate VLMs (fine-tuned Qwen3VL) and CV experts via LLaMA 3.1
• Created GeoMeasure by fusing SAM 2.1/3 with RemoteCLIP for numerical VQA
• Optimized a Tri-YOLO ensemble for scale variance (0.4m-30m GSD)
• Achieved <10s inference on A100 GPU, outperforming SOTA

• Built a fast sMRI preprocessing pipeline for ADNI scans
• Benchmarked CNN, 3D CNN, attention-based, and dual-branch architectures for Alzheimer's detection
• Achieved 85% accuracy, competitive with SOTA benchmarks
• Applied transfer learning for early-stage MCI detection

• Built a cost-effective 360-degree perception system that replaces LiDAR with monocular cameras
• Developed a Genetic Algorithm-based localization method from camera observations
• Fused it with odometry for real-time localization at 20 Hz
• Integrated YOLO11n + Geometric Ray Casting for 3D coordinates
• Achieved <10 cm positioning error without active depth sensors

• Trained YOLO11n with a Kinect depth camera for 3D ball tracking at 20 FPS
• Formulated trajectory prediction algorithm for interception point estimation
• Implemented custom velocity controller for goalkeeper movement
• Validated via custom 2D Python simulator before physical deployment

• Built a custom Tetris environment with Pygame rendering and video export utilities
• Implemented a Double DQN model and training pipeline for this environment