Hi, I'm Himon.

ttp

Nice to meet you. I like to research cool things in AI, and I sometimes write about things I find interesting.

About me.

I'm a part-time grad student and a full-time tired person. I'm working on my thesis on multi-agent planning and LLM metacognition. I also plan to start my PhD soon. I like to obsess over a lot of different things. I hope at least one of them will make me some money some day.

Experience.

I am working as a Graduate Instructor for the CS Department at UCCS, teaching Computer Organization and Assembly Language.

I also work as a student staff at the CS Department at UCCS.

This was my second summer working as an NSF REU Mentor. I had fun!

I was an Research Assistant (RA) at the Quantum AI and Security (QAS) Lab at UCCS. I contributed to the Q-Dev Project.

I have served as a Teaching Assistant (TA) for several courses in the past.

For more details, check out my resume.

My Research.

I like AI, the whole superset. I started how most people started back then, with neural networks for classification. Then I worked on image processing and disease detection with ResNets and graph neural networks. I then worked with Dr. Adham Atyabi at the NeuroCognition Laboratory (NCL) on brain-computer interface (BCI) and cognitive workload assessment. I was a Graduate Research Assistant (GRA) in the Quantum-Classical AI and Software Engineering (QAS) Lab with Dr. Armin Moin. I got to work on an energy-efficient AI project and submitted a poster to ICSE! Then I took a sharp turn and tried quantum computing in the Q-Dev Project working on quantum ML and quantum NLP.

I'm currently working with Dr. Jugal Kalita at the Language Information and Computation (LINC) Lab. I am and have been working on multi-agents, reinforcement learning, low-resource and underrepresented languages, knowledge distillation, multimodal language models, explainability and interpretability, among others.

Lately, I've been interested in post-transformer architectures, particularly MAMBA, JEPA, diffusion, rectified flow, linear attention, etc. Small language models, continuous learning, and emergent behaviors excite me a lot!.

What I typically use for my projects

I like simplicity and there are way too many tools out there. If I can oonga, then I boonga. I wrote this website in pure HTML and some splash of CSS.

I primarily work with Python for AI research and quick prototyping using Jupyter in VS Code or Google Colab. I like PyTorch. Tensorflow is fine too, but way too many version conflicts. I'm learning Go because it looks fun. I use LangChain, LangGraph, HuggingFace for LLM and agentic projects. I use Unsloth, Llama.cpp, Accelerate because they save me a ton of headaches.

I have used Cirq, Qiskit, Pennylane for quantum computing projects. I frequently use C, C++, Java, MATLAB for different stuff as needed.

I have worked with AWS products like SageMaker, EC2, S3, Braket, etc.

Hi, I'm Himon.

Updates.

About me.

Education.

Experience.

My Research.

What I typically use for my projects

Thanks for visiting!