At Amazon, I build and migrate fraud-enforcement workflows that process 2K+ TPS across 60+ fraud signals. My work has raised risk-evaluation accuracy above 97%, protected 370K+ customers, and prevented $4.2M+ in annual losses.
I improve the reliability and economics of Tier-1 backend systems, including cutting IMR costs by 50% through a rebuilt query engine and removing a partitioning hotspot across 5K+ RPS of insert traffic. I've also delivered 75+ REST APIs, load-balancer-backed services, and zero-downtime JDK 21 migrations.
I enjoy owning systems end to end, from design and AWS infrastructure through testing, deployment, and documentation. During my Amazon internship, I built an automated action-item tracker used across 30+ teams.
I also build practical AI tools, including AtlasBot, a GenAI support bot that reduced on-call reachouts by 70%, and Intent Lens, which reduced model-analysis time from two hours to 20 minutes. I mentor engineers and students in DSA and system design, and have mentored three SDE interns.
