• Developing and integrating on-device computer vision features using React Native, Google ML Kit (Face & Pose Detection), and native camera APIs for real-time user guidance.

• Building responsive UI components with TypeScript and React Native, ensuring cross-platform consistency and accessibility.

• Optimizing performance for low-latency inference by profiling render cycles, reducing bundle size, and implementing efficient state management.

• Designed and developed an offline NLP-based assistant using Python to automate information retrieval from user manuals.

• Integrated a pre-trained BERT model, fine-tuned it for domain-specific queries, and improved resolution accuracy to ~90%.

• Implemented a feedback-driven learning loop using user input logs stored in SQLite for continuous model improvement.

In my role at Scanpoint Geomatics Ltd, I spearheaded the “Agrivision” project, leading a team of interns to design a comprehensive ML-to-maps pipeline. I automated the ingestion and processing of substantial satellite imagery data, and engineered a backend service that published ML predictions to PostGIS. This significantly improved the efficiency of map layer delivery for field inspectors overseeing extensive agricultural areas.

• Designed and implemented IoT systems with sensor integration and wireless transmission.

• Developed microcontroller-based prototypes for environmental monitoring using DHT11 sensors.

• Participated in a capstone project demonstrating live data collection and display using LCD screens.