Hi! I'm a software engineer passionate about leveraging technology to create impactful, accessible, and thought-provoking experiences that foster connection and empower communities.
Currently working as a software engineer at JPMorgan Chase & Co.
EXPERIENCE
JPMorgan Chase & Co.
07.2025 - PresentSoftware Engineer
Working on the AI Technology Team
Bina Lab
02.2021 - 06.2025Research Lead
Developed the HurricaneVidNet dataset for disaster damage assessment using UAV-based video data. Modify and test video semantic segmentation models while overseeing a team of 12 researchers.
JPMorgan Chase & Co.
06.2024 - 08.2024Software Engineering Intern
Developed a metric tracking library for microservices and integrated it with AWS CloudWatch. Built a CloudWatch dashboard to enhance observability and led a global hackathon team, securing 3rd place out of 70+ teams.
JPMorgan Chase & Co.
06.2023 - 08.2023Software Engineering Fellow
Built a full-stack application using Spring Boot and React. Led a business expansion project for Citizens Trust Bank, delivering a strategic growth plan for Houston.
Ernst & Young
02.2023 - 06.2023Expedition EY Fellow
Chosen as one of 400 high-performing students worldwide for a program focused on technology, consulting, and cybersecurity advisory. Engaged in targeted workshops and networking with EY professionals to gain insights into analytics, consulting, and financial services.
Juni Learning
12.2022 - 06.2023Computer Science Instructor
Taught Java, Python, and JavaScript to K-12 students. Created personalized learning plans and maintained regular progress updates for students and parents.
SELECTED WORK
Kairn
Kairn is a local context governor and runtime for coding agents that keeps long sessions focused. It runs beside the agent and helps prevent repeated files, stale details, and irrelevant repo noise from taking over the working session.
Retrieval finds possible files and snippets. Kairn decides what the agent should actually see next based on the task, repo state, previous work, failures, and other execution signals.
Kairn supports Codex CLI session mode and MCP, stays local-first by design, and current public evidence shows around 40% median lower token use on clean successful rows plus 20/25 SWE-Pro rows passed with Codex + Kairn versus 15/25 for Codex alone.