DRIVING DETECTION MACHINE LEARNING ALGORITHM
- Developed a CNN + LTSM model for driver detection based on an international research papers.,
- Optimized model latency to achieve authentication in sub 100 seconds with an accuracy of 84%.,
- Transformed ML infrastructure to an ONNX for successful integration with AWS S3 buckets., Interfaced Lambda & EC2 instances with Dynamo cluster to automate model training.
Technologies - TensorFlow, AWS, panda
NAVPAL
- Led a three man team and worked diligently to meet various milestones.
- Implemented Dijkstra's shortest path algorithm, A search, for efficient multiple waypoint travel, Tasked with organizing API raw data into multiple data structures for an optimal user experience.
- Developed a GUI using OpenGL to allow users to search for street intersections and points of interest.
Technologies - C++, Vim, OpenGL
SVN ORM LAYER DEVELOPMENT
- Built ORM framework in Python to help with seamless conversion of in-memory types to and from storage types.
- Expanded functionality to incorporate support of complex data types such as datetime and coordinates.
- Built multi-threading support and exploited L1 cache to reduce processing speeds by tenfold
Technologies - Python, Django