My name is Rohan Mistry (@rmluck). I am an aspiring software engineer from Chino Hills, CA. Recently graduated from UC Irvine, with a Bachelor's Degree in Computer Science specializing in Intelligent Systems. Passionate about leveraging cutting-edge technologies to solve complex problems in the fields of Machine Learning, Web Development, Data Science, and Artificial Intelligence. Have gained experience in full-stack development, machine learning, information retrieval, data visualization, and backend data integration through internships, research projects, and academic coursework. Have strong analytical, programming, communication, and leadership skills, with a collaborative mindset and proactive learning approach, eager, to tackle new challenges in professional settings.

You can reach out to me via email at [email protected], Discord, Instagram, or LinkedIn!

• Participated in multi-faceted AI open source research project
• Focused on improving ability to model the world and predict outcomes of economic and policy interventions using advanced AI techniques like transformers and foundation models
• Contributed to Data Collection team by curating and processing large-scale time series data from open government and international sources including Federal Reserve Economic Data (FRED), NYC Open Data, etc.
• Contributed to Scaling Laws team by conducting analyses of research papers related to AI model behavior to inform performance optimization strategies across various computational resources
• Contributed to Model Architeceture Explorations team by investigating ideas for innovative architectures to enhance AI capabilities
• Contributed to Pretraining team by supporting development of large-scale pretraining pipeliens to prepare models for downstream tasks

• Played key role in streamlining product marketing processes through development and implementation of web scraping programs
• Automated data collection for industry research, significantly improving both the efficiency and accuracy of company's data gathering efforts
• Contributed to introduction of new products by creating and refining marketing collateral
• Reviewed web content for consistency and clarity
• Conducted thorough competitive analysis to better position products in market
• Actively managed inventory analysis using online databases such as SAP, ensuring that promotional activities and new production integrations were well-coordinated
• Contributed to client communications which drove forward sales initiatives
• Involved in analytics projects by researching industry trends, competitor performance, and market dynamics

• Oversaw the daily operations of a fleet of 50+ autonomous robots as part of an innovative food delivery platform
• Managed start-of-day and end-of-day tasks such as charging, cleaning, and performing routine maintenance to ensure that robots were fully functional and able to complete deliveries without disruption
• Worked closely with merchants to configure and program robots to carry out specific delivery tasks, ensuring that customer orders were fulfilled accurately and efficiently
• Troubleshooted during unexpected incidents and collaborated with maintenance team to facilitate infrastructure upgrades that improved overall efficiency of operations
• Maintained hub-based warehouse and conducted regular inventories

Below are the relevant courses I have taken at UC Irvine:

• Getting Started with AI using IBM Watson (IBM)
• Introduction to Artificial Intelligence (IBM)
• Introduction to Data Analytics (IBM)
• Introduction to Software Engineering (CodePath)
• Web Development (ICSSC Fellowship)

• Engineering Design and Development (2020-21)
• AP Computer Science Applications (2019-20)
• Civil Engineering and Architecture (2019-20)
• AP Computer Science Principles (2018-19)
• Introduction to Engineering Design (2017-18)

At UC Irvine, I was also involved in various organizations such as:

• Sports Business Association
• Data Science Club
• Artificial Intelligence Club
• Google Developer Student Club
• Cybersecurity Club
• Management Information Student Society
• Future Business Leaders of America
• Design Club
• Association of Computing Machinery

• Hosted guest speakers from sports business industry
• Networking opportunities and workshops to gain industry knowledge
• Private tours of nearby sports facilities
• Tracked organization member participation
• Conducted data analysis on social media engagement of organization's accounts, including Instagram
• Other web-related tasks

Python, Java, C, C++, HTML/CSS, JavaScript, SQL, PHP, R, React, R, AWS, Tableau, DBMS, Docker, TensorFlow, PyTorch

Full-stack web application designed and developed to simulate NFL draft scenarios, allowing users to control specific teams, make real-time draft selections, and view draft results dynamically. Built with a React (Vite) frontend with JavaScript and FastAPI backend with tools like SQLAlchemy and Pydantic, using PostgreSQL for database management. Deployed with Netlify and Render. Users can simulate NFL draft round-by-round, view and track draft picks live, select specific teams to control draft picks, and undo or trade picks. For non-user-controlled picks, CPU auto-selects from big board according to positional weight-based algorithm. Export options via PNG, CSV, and JSON. Responsive frontend web design using HTML and CSS optimized for desktop. CORS-configured secure API access.

Skills: Full Stack Web Development, Backend Data Management and Integration, UI/UX Web Design

Tech: Python, JavaScript, HTML, CSS, SQL

Frontend: React (Vite), React-Router

Backend: FastAPI (Python), SQLAlchemy, Pydantic

Database: PostgreSQL

Migrations: Alembic

Deployment: Netlify (frontend), Render (backend, database)

Bayesian regression model developed to predict Yelp restaurant star ratings based on review sentiment and restaurant popularity. Used review text sentiment analysis and business popularity metrics to analyze how customer perceptions and business popularity influence public ratings.

Skills: Machine Learning, Bayesian Linear Regression, Data Analysis, Data Visualization

Tech: Python, Jupyter Notebook

Pre-Processing and Data Analysis: pandas, numpy

Sentiment Analysis: nltk Vader

Built dynamic full stack architecture from scratch, including web application allowing customers to browse, search, and purchase films from large database consisting of over 15,000 films and 60,000 actors. Hosted application on AWS EC2 for scalable resources, integrated Tomcat, Maven, HTTPS, and MySQL. Developed ETL pipeline to parse large XML files to augment already large database. Constructed large frontend using HTML, CSS, JavaScript, jQuery, and Ajax. Implemented over 20 features within fully-functional application including importing catalogue of thousands of films, full-text search with autocomplete and sorting backed by cache and JDBC, session-based cart checkout, secure login using SHA256 password encryption hashing, bot detection using reCAPTCHA, and protection against SQL injection attacks via PreparedStatements. Improved website performance 30% by implementing optimization techniques such as MySQL connection pooling, MySQL replication, and Apache load balancing. Deployed Docker-containerized version of project on Kubernetes cluster spanning multiple AWS instances. Leveraged JMeter to analyze application performance.

Skills: Full Stack Web Development, Backend Data Management and Data Integration, UI/UX Web Design

Tech: Java, SQL, JavaScript, HTML, CSS, Ajax, jQuery

Designed and implemented multi-agent reinforcement learning (MARL) framework to optimize traffic signal control in urban environments. Focused on leveraging advanced algorithms like multi-agent proximal policy optimization (MAPPO) to dynamically adjust signal timings based on real-time traffic data, reducing congestion, minimizing vehicle delays, and increasing overall traffic throughput. System treats each intersection as independent agent capable of coordinating actions with neighboring agents while utilizing centralized training and decentralized execution. Implemented dynamic reward functions to encourage smooth traffic flow, integrated CityFlow simulation environments for large-scale testing, and used real-world datasets such as UC Irvine Traffic Flow Forecasting dataset for model evaluation. Implemented features such as detailed state representations and action spaces to manage signal phase configurations. Leveraged Ray's RLLib library to train and tune model on training simulations and improve model. Model achieved significant improvements in metrics such as mean velocity, halting duration, and lane occupancy, demonstrating scalability and adaptability in complex traffic networks.

Skills: Artificial Intelligence, Machine Learning, Reinforcement Learning, Data Analysis

Tech: Python

🏆 Awarded "Best Analysis of Airbnb Market in Dublin [Sponsored by StrataScratch]" Award at UC Irvine Datathon 2024. Conducted in-depth exploratory data analysis using StrataScratch datasets. Collaborated with team to visualize insights on market supply and demand, focusing on elevating Airbnb user experience for guests and hosts. Leveraged tools such as DeepNote, Alteryx Designer, and Python libraries (Pandas, Matplotlib, Seaborn, NumPy, Scikit-Learn). Developed and tested decision tree classifier machine learning model using Python that achieved 89.35% accuracy rate leveraging tools like DeepNote and Alteryx Designer.

Skills: Data Analysis, Data Visualization, Machine Learning

Tech: Python, Jupyter Notebook

Data Analysis: pandas, numpy, Alteryx Designer

Machine Learning: scikit-learn

Visualization: seaborn, matplotlib

Presentation: DeepNote, Google Slides

Collaboration: Google Workspace

Web application developed using Python and TensorFlow/Keras to classify images from the Fashion MNIST dataset which consists of 70,000 images of various items of clothing. Users can upload grayscale 28x28 images to receive predictions from several employed scikit-learn machine learning models including logistic regression, K-nearest neighbors, feed-forward neural networks, and convolutional neural networks. The app also supports evaluation on the test set with detailed visualizations.

Skills: Machine Learning, Image Detection, Data Visualization

Tech: Python

Frontend: Streamlit

Machine Learning: TensorFlow/Keras, scikit-learn

Data Processing: numpy, PIL (Pillow)

Visualization: matplotlib, seaborn, pandas

Deployment: Streamlit Community Cloud

Web application designed and developed using Python libraries to process and analyze Netflix viewing history, deployed with Streamlit. Generated visualizations and insights regarding viewing patterns based on frequency, duration, and location, providing comprehensive overview of user behavior.

Skills: Web Development, Data Analysis, Statistics, Data Visualization

Tech: Python

Frontend: Streamlit

Data Analysis: pandas, numpy

Visualization: matplotlib, seaborn

Deployment: Streamlit Community Cloud

Developed a Checkers AI agent using Python which could solve a Checkers game and win over 50% of games against an Average AI model. The agent could read percepts and act rationally based on the opposing player's moves. Implemented Monte Carlo tree search.

Skills: Python, Artificial Intelligence

Implemented and designed a customized web search engine and crawler that handled 50,000+ documents under operational constraints with a query response time under 100ms. Incorporated partial indexing and used a Porter stemmer to improve textual matches. The scraper parsed web responses and extracted critical information for a detailed report. Improved ranking accuracy by 75% for a subset of UCI web domain pages through tf-idf scoring and cosine similarity alongside weighted HTML tags. Used sim-hashing to detect and eliminate near-duplicate pages. Built and incorporated a tokenizer that processes text and can output word frequencies of a page and shared word frequencies between two pages.

Skills: Beautiful Soup, Search Engine Optimization, Web Scraping, Python, HTML

Designed fully-functioning Columns match-three puzzle video game with grid-field of cells and cooled jewels that fall, freeze, match, rotate, shift, and disappear throughout gameplay. PyGame implemented to build playable version of game.

Skills: Pygame, Python

Created air quality analysis program that takes location input and Air Quality Index threshold and connects to PurpleAir web API for air quality monitoring sensor data to determine variable number of nearby locations with poorest air quality. Nominatim API, forward geocoding, and reverse geocoding methods used to calculate nearest locations.

Skills: Data Analysis, API, Python, Geocoding

Capstone group project involving a complete engineering research and design process. Investigated and researched solar ultraviolet radiation issues relating to driving. Documented and analyzed prior solution attempts and patents on the market for automobile sun visors that block sun radiation. Brainstormed potential design concepts and generated sketches and 3D-models, concluded by printing and constructing a physical prototype of a car visor. Presented to Civil Engineering and Architecture class.

Skills: Engineering Research, Market Research, Prototyping, Product Design, 3D Modeling

Prior to entering college, I was a student at Ruben S. Ayala High, part of their Academy of Engineering and Computer Sciences for four years. Co-founded the Ayala VEX Robotics Club and Ayala Information & Computer Science Club. For the former, I was treasurer for three years, managing the club's budget, scheduling fundraisers and purchasing necessary robotics parts and equipment. I was also the primary programmer (C++) for my individual robotics team for four years.