Senior Software Engineer · IoT Products and Platforms
Jun 2018 – Aug 2020 · Bangalore, India
- Senior Software Engineer in the IoT Products and Platforms division at Samsung Research India (SRI-B), Bangalore, the largest Samsung R&D center outside South Korea. Worked across research and product engineering on connected-device experiences, including the Galaxy Home smart speaker, the SmartThings services platform, and on-device computer vision for surveillance cameras. Cleared Samsung’s internal Software Competency Test (Professional level) within the first year.
Anomaly Detection Research
- Lead author on “A Multi-Input Neural Network with Dense Flow and Spatio-Temporal Features for Anomaly Detection”, presented at the 15th International Conference on Information Processing (IEEE ICinPro 2019) in Bangalore. The paper has since been cited in further peer-reviewed work, including a 2023 Multimedia Tools and Applications article on data-driven road accident detection.
- The technical contribution is a neural network that fuses spatio-temporal features from a 3D convolutional network (C3D) with dense optical-flow features capturing motion. Earlier approaches typically used one or the other; combining them allowed the system to both detect anomalies and localize when in the video they occurred, with lower false-alarm rates on motion-heavy normal footage than the baselines evaluated against. Also contributed a standardized dataset for further research in the area.
Galaxy Home Audio Monitoring
- Built an emergency home audio monitoring engine to detect common household sounds (glass breaking, smoke alarms, and similar) for use as ambient safety signals. Integrated into the Galaxy Home smart speaker program.
- Received a Samsung spot award for the work and presented the innovation at an internal workshop in Korea.
WebRTC SDK for IoT Devices
- Developed an SDK for IoT devices to enable efficient peer-to-peer real-time communication using WebRTC, reducing latency, server costs, and privacy concerns relative to server-relayed approaches. Integrated into devices including surveillance cameras and robot vacuum cleaners.