Anujay Ghosh - Portfolio

About me

Welcome to my portfolio! I am an AI and Software Developer with a Master's degree in Computer Science from Stony Brook University - SUNY. I enjoy turning complex problems into simple, beautiful, and intuitive designs.

I have experience in building AI-focused applications and a strong background in software development, utilizing modern technologies to create user-friendly and visually appealing solutions.

What I'm Doing

AI/ML

Cutting-edge AI/ML solutions crafted for optimal performance.
NLP and LLMs

Innovative NLP and LLMs solutions designed to improve user interactions.
Product Development

Innovative product development for Windows and Android.
Software Development

Scalable full-stack software development to meet business needs.

Education

Stony Brook University (State University of New York)
Jan 2023 - Dec 2024
Master of Science, Computer Science
• Coursework: M.S. Thesis in Accessibility using AI(LLMs), Computer Vision, Data Science, Visualization, Databases, Probability and Statistics
University of Mumbai
Aug 2016 - Nov 2020
Bachelor of Engineering, Computer Engineering
• Coursework: Machine Learning, AI, NLP, Operating Systems, Computer Networks, Software Engineering

Experience

AI and Software Engineer Consultant
Stealth AI Startup, Feb 2025 - May 2025

Designed and implemented secure, scalable backend APIs using Node.js, Express, and TypeScript, enabling OAuth2 authentication and Reddit API integration supporting over 10,000 automated actions per day. Engineered end-to-end Reddit automation: implemented endpoints for post/comment creation, subreddit search, and analysis.

Developed advanced AI-powered features by integrating OpenAI and Anthropic APIs for automated content generation, and intelligent workflow suggestions, automating content generation and analysis for 1,000 workflows, reducing manual moderation time by 60%. Resolved critical bugs and shipped high-impact features such as AI-generated Reddit comments, and robust error handling, decreasing production error rates by 70%.

Enhanced security and privacy by encrypting sensitive data, secure credential storage, and comprehensive input validation, resulting in zero security incidents. Optimized application performance through response caching and batch processing, reducing API response times by 50% and supporting 5x more concurrent users.
Graduate AI Research Assistant
Dept. of Computer Science - SUNY at Stony Brook, Aug 2023 - Dec 2024

Developed Savant, an assistive technology powered by large language models (LLMs) to enable uniform computer interaction for blind users across diverse applications, automating screen reader actions through MS UI Automation and advanced NLP techniques in Python, achieving 90% accuracy through natural language commands.

Enhanced LLMs by implementing fine-tuning and few-shot learning to interpret natural language commands, automating the correct sequence of screen reader actions in an application-agnostic solution, resulting in improved user satisfaction and efficiency.

Conducted user studies with 11 blind participants, demonstrating significant improvements in interaction efficiency (4x faster) and usability (3x higher SEQ score) compared to standard screen readers.
Software Engineer
Accenture, Jan 2021 - Jan 2023

Improved Cigna's Enrollment strategy backend with Java Spring Boot, AWS (S3, CloudWatch, VPCs), and REST APIs. Achieved a 30% scalability boost in application development and 40% faster response times.

Reduced client's AWS costs by 30% while maintaining product quality through EC2 reserved instances, using Python 3 for storage, CloudWatch for monitoring, and VPCs for network isolation and reliability, while also leveraging AWS Lambdas.

Collaborated in a 5-member cross-functional Agile development team, utilizing Java, Python, RESTful APIs, SQL, and AWS services. Achieved a 15% productivity boost and on-time, on-budget delivery of 4 major projects.