Anshul Doshi

• Built a TCP Client/Server using system calls supporting uploading/downloading, deleting, and modifying files on a server.
• Implemented event-driven programming and non-blocking I/O to efficiently handle multiple clients on one thread.
• Used: C programming language, Socket (IP/TCP) Programming, epoll for event-driven programming

• Wrote a multi-threaded version of the make tool that utilizes pthreads and mutex locks to parallelize a make command.
• Identified file dependencies by building a tree of all all relevant files using this structure to properly schedule commands so
that all relevant files are up to date. Checked file timestamps to avoid redundant compilations.
• Used: C programming language, pthread library and mutex locks for multithreading

• Developed a shell capable of running programs in the foreground or background, spawning multiple processes, and cleaning up resources when processes terminate to prevent the creation of zombies.
• Supports all standard bash commands (ls, cd, etc.) and built-in commands (kill process, stop process, continue process).
• Used: C programming language, Linux system calls fork(), exec(), wait() to handle multiple processes

• Wrote a memory allocator using sbrk to manually control heap size capable of supporting malloc, calloc, realloc, and free.
• Implemented First-Fit algorithm and maintained a linked list of free memory blocks for quick retrieval of free memory.
• Utilized memory merging to consolidate free, adjacent memory blocks and prevent internal fragmentation.
• Used: C programming language, Linux system call sbrk() for memory management

• Designed an app in Xcode to calculate the difference between the time experienced by the user and the time experienced by one at rest using the principles of Special Relativity.
• Obtained velocity and location data using the CoreLocation library in order to provide real-time data to the user.
• Used: Swift programming language, Xcode, CoreLocation library

• Built a CNN on Keras to identify the orientation of an object using an image of the object as input and a vector as output.
• Achieved 96% model accuracy after optimizing variables such as the number of hidden layers, batch size, and applying regularization techniques to maximize model accuracy and minimize effects of overfitting.
• Used: Python, Numpy for computations on large pixel arrays, Matplotlib for visualizing results, Keras for ML libraries

• Designed a Digital Audio Workstation that provides musical suggestions to the user to aid in the process of music composition.
• Parsed user-selected MIDI files using Python in order to build 1st and 2nd order Markov Chains. Utilized these models to suggest the 3 most probable notes given a previous one.
• Designed a UI to create songs and visualize the user’s composition as it develops; also allows for preview listening.
• Used: Python, NumPy, Music21 library for MIDI parsing, PyQt5… Show more

• Designed a Digital Audio Workstation that provides musical suggestions to the user to aid in the process of music composition.
• Parsed user-selected MIDI files using Python in order to build 1st and 2nd order Markov Chains. Utilized these models to suggest the 3 most probable notes given a previous one.
• Designed a UI to create songs and visualize the user’s composition as it develops; also allows for preview listening.
• Used: Python, NumPy, Music21 library for MIDI parsing, PyQt5 for GUI design. Show less