Software Developer

About Me

Hello, I am Sam Perelgut, a software developer with a specialization in machine learning. I am a relatively recent graduate of the University of California, Irvine, where I earned my M.S. in Computer Science. Prior to UCI, I attended the University of California, Davis, where I double majored in Computer Science and Cognitive Science. I am currently working in game development at a new start-up called NotHalfBadGames as a contract software developer. I am passionate about AI and its applications in a variety of fields. I believe it can be used to change the world, and I would like to be part of that change.

Technical Skills: Python, C/C++, Tensorflow/Keras, PyTorch, SQL, GDScript

Education

• M.S., Computer Science

  University of California Irvine (December 2022)

• B.S, Computer Science

  University of California Davis (June 2021)

• B.S., Cognitive Science

  University of California Davis (June 2021)

Work Experience

Software Engineer @ NotHalfBadGames (Feb 2023 - Present)

• Significant contributions as a software engineer for an unreleased global economics-based strategy video game which overall led to a 50% increase in time saved for the team’s development lifecycle
• Wrote and debugged 100% of the code for pop-ups, menus, tool bars, data clustering, mini-map, material shaders, external data source processing, interface/backend integration, save/load states, demos, microenvironment physics simulations, Voronoi diagrams, Procedural Terrain Generation.
• Optimizations to physics engine, code simplification, and utilization of advanced run-time algorithms which increased games frame rate from 25 to 90 frames per second on company’s benchmarks
• Extensive data analysis focusing on visualization and manipulation of large amounts of CSV data in order to incorporate accurate geographical information into game environment.
• All tasks performed utilizing advanced data structures, extensive object-oriented programming, and Git with programming in Python, GDScript, and QGIS.

Student Teaching Assistant @ University of California Davis (Jan 2021 - june 2021)

• Worked as a student teaching assistant for an Introductory C programming Course
• Taught students object-oriented programming in Python, helped organize material for lectures, and assisted students in debugging code

Projects

Shipping Game Simulator

This is the project I have been working on for my current job postion from Feb 2023 to the present. The animatiom below focuses on functionality and showcases some key features such as:

• Menu and user UI,
• Navigation Mesh allowing users to draw their own routes with AI assistance
• Animations including movem,ent of ships between ports,
• User navigation of map with synchronized updates to the mini-map.

In addition to core gameplay features, I’ve focused on enhancing the visual polish of the game as well through the following: Shaders: Utilized advanced shader techniques to create visually appealing effects,

• including realistic lighting and shadow
• seamless textures
• realisitic terrain
• flowing waves which crash on the shoreline

Heart Attack Detection Website

GitHub Repo and Report and Results

• Served as team leader and full stack developer for creation of a website which would classify whether users are at risk of a heart attack
• Performed data analysis, Front-end development, back-end development and integration of the two using PyTorch, React, and Flask
• Supervised Learning Classifier utilizing neural network over Big Data

Sokoban Solver

• Developed an artificial intelligence program that uses Reinforcement Learning/Q-Learning to solve Sokoban boards in Python.
• Utilized the A* algorithm and softlock detection to increase performance

Seq2Seq Title Generation

GitHub Repo and Report and Results

• Developed a Natural Language Processing program using PyTorch to generate titles for forum posts.
• Used Seq2Seq models including RNN model, BART Transformer model, and BigBird-Pegasus Transformer model.
• Utilized attention model and word embeddings to build decoder, evaluated using BLEU

Ensemble Classifier

GitHub Repo

• Developed an Ensemble classifier using Scikit-learn to classify images in the Fashion-MNIST dataset
• Utilized Gradient Boosting, Support Vector Machine, Random Forrest and Convolutional Neural Network to build an Ensemble classifier