About Me

Working on the modernization and development of computer-vision systems, AI infrastructure, and video-analytics platforms for real-world deployments.

▪ Design and develop production computer-vision and video-analytics systems.

▪ Build multi-camera analytics pipelines for detection, tracking, re-identification, heatmaps, dwell-time analysis, occupancy, zone transitions, and temporal analytics.

▪ Design edge-first and centralized inference architectures using NVIDIA GPU platforms.

▪ Work with NVIDIA DeepStream, GStreamer, TensorRT, ONNX, RTSP, and GPU video encode/decode pipelines.

▪ Design backend and analytics infrastructure for high-volume event and telemetry data.

▪ Work on deployment architecture spanning edge devices, on-premise GPU servers, and cloud infrastructure.

▪ Contribute to camera, GPU, compute, networking, and storage architecture for production deployments.

▪ Develop infrastructure and deployment practices using Docker, Terraform, Ansible, monitoring, automated backups, and secure remote-access systems.

▪ Collaborate on technical architecture, infrastructure planning, hardware evaluation, and production deployment strategy.

Worked across computer vision, geospatial AI, multimodal learning, production inference infrastructure, backend systems, and agentic AI products.

Geospatial AI & Digitalization of Quality

▪ Developed multimodal AI systems combining imagery, sensor data, and structured features for predictive quality assessment.

▪ Worked across the complete ML lifecycle from literature review and prototyping to evaluation, optimization, backend integration, and deployment.

▪ Designed multispectral and multitemporal satellite and UAV processing pipelines for crop-health monitoring and environmental analysis.

▪ Developed change-detection systems for vegetation, soil, and land-use analysis.

▪ Worked on GNSS-based autonomous UAV path planning for field coverage.

▪ Developed production inference systems using ONNX Runtime and Triton Inference Server.

▪ Built supporting FastAPI services and asynchronous inference workflows.

Agentic AI

▪ Worked on the architecture of an agent-based software engineering platform driven by natural-language requirements.

▪ Developed multi-agent orchestration, configurable LLM routing, retrieval-backed memory, isolated execution, and automated code-review workflows.

▪ Designed structured interfaces between AI agents and execution systems.

Recognition

▪ A satellite-image change-detection system developed during this work was shortlisted in the NCIIPC AI Grand Challenge.

Developed computer-vision, edge-AI, and backend systems for drone-based inspection and remote-sensing applications across agriculture, conservation, solar infrastructure, and warehouse automation.

Solar Plant Management & Inspection

▪ Developed a solar-panel detection model achieving 97.2% bounding-box accuracy on RGB and thermal imagery.

▪ Developed a 12-class panel-condition classifier achieving 91.3% accuracy for solar-panel fault identification.

▪ Used multi-view geometry and triangulation to associate detected faults with their physical position within solar installations.

▪ Built backend APIs, inference scheduling, and task-management infrastructure supporting CV inference workflows.

▪ Deployed models to edge platforms including NVIDIA Jetson.

Aerial Rhino Interception System

▪ Developed computer-vision systems for detecting humans and animals in thermal imagery.

▪ Built sensor-processing pipelines using IMU, acceleration, vibration, GPS, and drone telemetry.

▪ Geo-tagged detected objects using UAV telemetry to support location-aware alerting.

Crop Health Monitoring

▪ Developed weed detection and segmentation techniques using NIR imagery.

▪ Investigated spectral-index-based soil-water assessment.

▪ Developed crop-health prototypes combining RGB and NDVI imagery.

Warehouse Inventory Tracking

▪ Designed a camera-based warehouse positioning concept for inventory tracking.

▪ Evaluated deployment cost and technical constraints and pivoted the system toward QR-based positioning.

▪ Delivered a QR tracking pilot accepted for client rollout.

Backend & Deployment

▪ Developed Flask/FastAPI backend services supporting ML inference.

▪ Built inference scheduling and task-processing pipelines.

▪ Deployed services on GCP through CI/CD workflows.

▪ Used ROS and Gazebo simulation for UAV and perception experimentation.

Artificial intelligence internship contributing to high-level machine learning and computer vision work.