Jinu Nyachhyon
AI Researcher | ML Engineer
About Me
I am a research-focused AI engineer working on LLM-driven agents and program reasoning. At Manifold Research, I develop and benchmark software control agents, building trajectory generation and stress-testing pipelines to evaluate robustness under distribution shifts. At Insyde AI, I design LLM-aided agent pipelines for multi-scenario reasoning, tool use, and end-to-end automation, achieving significant efficiency gains.
My research interests lie in LLM-aided program analysis, trustworthy and aligned AI, and secure multi-agent reasoning systems, with a focus on building scalable, interpretable, and robust agent frameworks.
Research Interests
Natural Language Processing
Large language models, multimodal learning, and cross-lingual transfer.
AI Alignment
Fairness, interpretability, and robustness of machine learning models.
LLM-Aided Agents
LLM-driven agents, multi-agent reasoning, and end-to-end task automation.
Computer Vision
Object detection, image segmentation, and visual reasoning with limited supervision.
Education
B.E. in Computer Engineering
Institute of Engineering, Tribhuvan University, 2018 - 2023
Research Area: Recommender System and Machine Learning
Advisor: Prof. Subarna Shakya
Recent Blog
December 29, 2023
Machine Learning Fundamentals: Part 5 - Advanced Topics and Real-World Applications
Final part covering MLOps, ethical considerations, and future trends in machine learning.
December 22, 2023
Machine Learning Fundamentals: Part 4 - Model Evaluation and Validation
Comprehensive guide to model evaluation techniques and validation strategies.
November 15, 2023
Understanding Transformers: Part 3 - The Complete Architecture
Exploring the complete Transformer architecture including encoders, decoders, and all components.
October 12, 2023
Advances in Multimodal Learning
Exploring recent advances in multimodal learning and AI systems understanding multiple forms of data.