Software Engineer @ AWS GuardDuty | AI Researcher | NLP Expert
I work on large-scale threat detection systems at AWS and explore AI safety, multi-agent systems, and robust machine learning through research and writing. My goal is to develop systematic thinking and contribute to AI research through rigorous technical analysis.
Building the future of AI through research and engineering
I am a Software Engineer at AWS GuardDuty with a deep passion for advancing Artificial Intelligence, particularly in Natural Language Processing (NLP). My work focuses on addressing the limitations of current AI systems in areas such as robustness, generalization, and ethical impact.
Through projects involving benchmark development, data quality improvement, and efficient training strategies, I aim to create AI models that are not only technically advanced but also socially responsible. I bring over six years of industry experience and three years of research to my work, where I explore cutting-edge solutions in machine learning, threat detection, and adaptive problem-solving.
My research interests include multi-agent systems, AI safety, and human-AI collaboration. I believe that by combining human "slow thinking" with AI's "fast thinking," we can achieve extraordinary outcomes while maintaining ethical considerations and positive impacts on human psychology.
Professional journey in AI and software engineering
Contributing to scalable and secure tier-1 services in the AWS GuardDuty team, working on large-scale threat detection systems and machine learning applications. Focus on building robust, high-performance systems that protect AWS customers from security threats.
Promoted multilingual AI as part of Project Aya, focusing on the Panjabi language and contributing to inclusive AI development. Worked on improving language representation in large language models.
Conducted research on numerical reasoning benchmarks for NLP under Dr. Chitta Baral. Published multiple papers in top-tier conferences including ACL and EACL. Focused on benchmark quality, data artifacts, and mathematical reasoning in AI systems.
Worked on IPP implementation, database migration, and machine learning prototypes in enterprise systems. Gained experience in large-scale system architecture and enterprise software development.
Research contributions to AI and machine learning
EACL 2023
Introduces VAIDA, a system that improves NLP benchmark quality through real-time visual feedback and metric-based recommendations, achieving a 45.8% reduction in artifact levels.
ACL 2022
A comprehensive benchmark that tests models' mathematical reasoning skills, revealing gaps in model understanding and pushing improvements in fundamental math reasoning.
ACL 2022
Investigates how generalization techniques improve NLP model performance in out-of-domain tasks, emphasizing the trade-off between generalization and robustness.
ICML Workshop 2020
Presents a framework to assess benchmark quality by identifying and correcting artifacts in datasets, promoting more robust and reliable evaluations.
SUSTAINLP Workshop 2020
Explores the efficiency of training NLP models using smaller, well-targeted datasets, promoting a more sustainable approach to AI development.
Journal of Intelligent & Fuzzy Systems 2016
Proposes a novel approach to traffic state detection using acoustic data collected from smartphones, offering a scalable solution to urban traffic monitoring.
Technical writing on AI safety and research
2024
Examining governance frameworks for AI systems that can evolve and adapt, ensuring human oversight and control.
2024
Analysis of how human "slow thinking" can complement AI "fast thinking" for optimal collaborative outcomes.
2024
Exploring the evolution from single large models to compound AI systems and their architectural implications.
2024
Personal experience working on multilingual AI development and language preservation through Cohere's Project Aya.
2024
Insights and lessons learned from enterprise software development and business operations during my time at Unisys.
Let's connect and collaborate
I'm always interested in discussing AI research, collaboration opportunities, and innovative projects. Feel free to reach out!
Currently exploring:
Open to research collaborations, speaking opportunities, and consulting projects in these areas.