Graduated with M.S. in Computer Science from Clemson University (GPA: 3.9/4.0).
Krishna Panthi light_mode dark_mode
Clemson, SC 29631, USA
I'm an MS student in Computer Science at Clemson University, working on reinforcement learning, time-series prediction, simulation environments and diffusion models. Before this, I was a Software Engineer at MutualArt. I completed my BE in Computer Engineering at Pulchowk Campus, Tribhuvan University.
In my free time, I enjoy playing tennis, working out, hiking, and making people laugh whenever I can.
Reach me out at
💼 Experience
Research Experience
- Developed and optimized AquaCrop-Richards, an open-source crop simulation model that integrates the Richards equation into AquaCrop, implementing a 1-D finite-difference solver with modified Picard iteration and Anderson Acceleration for faster convergence. Verified the implementation with field data, resulting in a peer-reviewed publication.
- Implemented and compared deep reinforcement learning algorithms (PPO, DQN, SAC) to study how the learning environment affects optimized irrigation policies, using AquaCrop-Richards as the simulation environment — work submitted as a first-author paper currently under review.
- Contributed to implementing the HydroQuantum Python library to do quantum machine learning on hydrology-related data. The associated paper has been published in Environmental Modelling & Software.
- Processed and analyzed large-scale graph data (38M+ edges) using Boost C++ to map upstream/downstream connectivity of U.S. river stations across continental flowlines.
- Took lead to create teaching materials presented at WaterSoftHack 2025, and deployed and managed GPU nodes on an HPC cluster using Kubernetes for WaterSoftHack 2026.
Professional Experience
- Developed, owned, and maintained full-stack applications for communication and sales serving thousands of users worldwide, using Vue.js and .NET Core with SQL Server, GraphQL, Mixpanel, and Elasticsearch.
- Built a marketing platform for drafting and scheduling mass-email campaigns across continents and time zones, reaching 200k+ recipients per event and cutting campaign creation time from hours to minutes.
- Built image- and PDF-processing workflows using Magick.NET and iTextSharp for art catalogs and invoices.
- Containerized legacy .NET applications with Docker, reducing deployment time.
- Implemented pixel-tracking tools to monitor email engagement and wrote unit tests with NUnit to validate business logic.
- Migrated features from a WPF desktop application to a web application by implementing custom Angular UI components, using Material UI and Telerik to build dashboards and other features.
- Optimized SQL queries and implemented Redis caching to improve data retrieval performance.
🎓 Education
Documents
News
Defended M.S. thesis: "Coupling the Richards Equation with AquaCrop and Deep Reinforcement Learning for Irrigation Scheduling Optimization."
Won Most Viable Product award at LPL Financial's 2026 Multi-University AI Hackathon (36 teams) for building a RAG-based compliance verification system using AWS Textract, Bedrock, DynamoDB, and Lambda.
Book chapter accepted in GeoHorizon (co-authored).
Paper on quantum machine learning for hydrology published in Environmental Modelling & Software (co-authored).