Obed Junias

I’m a second-year Master’s student in Computer Science at the University of Colorado Boulder, advised by Dr. Maria L. Pacheco in the BLAST Lab. My work sits at the intersection of interpretable reasoning and responsible AI.

My current research focuses on developing interpretable reasoning systems and benchmarks for commonsense and logical inference in natural language processing. I am exploring entailment tree-based frameworks as a way to make machine reasoning more transparent, structured, and logically grounded. In parallel, I work with Dr. Theodora Chaspari on bias detection and fairness evaluation in LLMs, particularly within the mental health domain—work that reflects my broader commitment to building AI systems that remain trustworthy and equitable across diverse populations.

Looking ahead, I am interested in exploring how structured reasoning and social or moral alignment can be incorporated into foundation model pretraining, and how models internalize, perceive, and compare these signals during inference and generation. I also aim to investigate reflective post-training frameworks, including reinforcement learning (RL)-based approaches, that encourage models to reflect, revise, and align with human values.

From a systems perspective, I am interested in how these mechanisms can be implemented and evaluated at scale in open, transparent, and sustainable model ecosystems.

Research Interests

• Natural Language Processing, Understanding, and Reasoning: Commonsense reasoning, neuro-symbolic methods for interpretable NLP
• Responsible and Interpretable AI for Social Good: Fairness and bias mitigation, social and moral alignment in LLMs, ethical and equitable development of AI
• Agentic and Reflective Systems: Self-reflective and introspective agents, multi-agent collaboration
• MLSys and Reinforcement Learning: Open, transparent, and sustainable model ecosystems, reasoning-aware pretraining, reflective post-training, and reinforcement learning for alignment

I’m actively seeking opportunities in NLP and related areas.
Feel free to reach out for research collaborations or other opportunities.