Evan’s AI Engineer Portfolio
Evan A. Khanna
I build AI systems that work in teams, not silos. As a Computer Science Honors student at James Madison University and founder of Dareshift LLC, I focus on multi-agent architecture, the idea that a suite of specialized AI agents, used deliberately, outperforms any single model used as a crutch.
That thinking shows up across everything I build: Dareshift, a medication management platform now in active beta; AI Debate Lab, a multi-agent debate application where independent AI agents argue, moderate, and judge each other; and ongoing research into how multiple language models reason together, and where that reasoning can drift.
I’m also Education Chair of JMU’s Madison AI Club, where I help other students build real judgment about when AI helps and when it doesn’t, because the tools only matter if you know how to use them.
I’m always open to collaborating with founders, researchers, and fellow builders thinking about what comes after single-model AI.
My Experience
Applied Computer Science in Practice
Across academic work, student leadership, and independent technical projects, I focus on turning computer science concepts into real, usable systems. Whether meeting with university leadership to represent student needs, building experimental AI applications, or optimizing web experiences, I’ve learned to combine technical problem-solving with clear communication and practical impact.
These experiences have strengthened my ability to take ideas from early concepts to working prototypes, collaborate with others, and continuously iterate based on feedback.
I’m especially interested in applying AI and software engineering to build tools that are both technically sound and genuinely useful to the people who use them.
Dareshift LLC
Founder and CEO
– Founded software company in 2017, shipped multiple titles on itch.io spanning game development and AI applications
Present
Madison AI Club (MAIC)
Education Chair
– Collaborating with club leadership to develop curriculum and programming for upcoming year
– Holding lessons on different AI concepts during meetings and helping others set up workshops on varying skills
Present
JMU Office of Student Awards, Initiatives, and Research
Research Assistant
– Selected for JMU’s competitive First-Year Research Experience (FYRE) program
– Contributed to the Voice Video Deepfake Detection research project by writing final research report and presentation poster covering MobileViT, EfficientNet-B0, and a WebRTC-based real-time inference pipeline
2026
The Johns Hopkins University Applied Physics Laboratory (APL)
ASPIRE Intern
– Programmed a fully playable space game in Unity/C# in a team of 3 and presented at a student showcase
2024
Use AI as a booster, not a crutch.
my Skills
I build practical AI-driven products from the ground up, combining machine learning, software engineering, and product thinking to turn ideas into real, usable tools.
- AI & machine learning
- software & product development
- small business management
AI & machine learning
- Prompt Engineering
- OpenAI API
- OpenRouter API
- Gemini API
- LLMOps
- Hugging Face
- Claude Code
Software & product development
-
Java, C#, Python (TensorFlow, scikit-learn)
-
Unity
-
Unreal Engine 5
-
Git/Git LFS
-
GitHub
Entrepreneurship
- Product Strategy
- Product Validation
- User Acquisition
- Growth & Search Engine Optimization (SEO)
- Analytics & Metrics
Awards & Education
BUILD.org
BUILDFest DMV 2nd Place
Pitched Dareshift LLC in a 5-minute pitch to 4 business executives from Fortune 500 companies and an audience
2025
BUILD.org
BUILD Hackathon winner
Created an AI therapy chatbot in addition to leading a team to create other marketing material, all within 30 minutes
2025
Microsoft
Foundational C# with Microsoft Developer Certification (May 2025)
Demonstrated proficiency in core C# programming concepts, validated through Microsoft’s official developer certification exam.
2025
James Madison University (JMU)
Bachelor of Science
Computer Science major, Honors Interdisciplinary Studies minor, Entrepreneurship minor
2025 – 2029
I’ve Worked With
2026
AI Debate Lab
A fully deployed, multi-agent AI application where two AI bots debate any topic you throw at them.
Built with Flask and the OpenRouter API, AI Debate Lab supports a near-unlimited number of AI agents, each with assignable, custom personas, and streams responses in real time. The platform follows an Oxford-style debate format with structured rebuttals — a moderator agent guides the conversation, and a judge agent delivers a final verdict. For supported models, agents can pull in live web search for their arguments. This project is my clearest proof point for a broader idea I’m exploring: that a suite of coordinated AI agents, used deliberately, outperforms any single model working alone.
You can download a formatted PDF transcript of every debate, printable and shareable.
2026
CarbonCoach
CarbonCoach is a machine learning powered web app that predicts a user’s carbon footprint and provides personalized strategies to reduce it.
Built with a partner at HooHacks (UVA), CarbonCoach uses a Random Forest regression pipeline to predict carbon footprint from user inputs like travel habits and waste disposal, identifying the behavioral features that matter most. An integrated Gemini API layer then turns those predictions into targeted, actionable reduction strategies tied to the user’s actual habits, not generic advice.