Inside JPMorgan Chase
I joined as a Global Finance and Business Management Summer Analyst and went straight into product mode.
Find the people closest to the problem, then build the thing that removes their friction.
Seven product leaders in seven days. I coffee-chatted seven product leaders across the firm in my first week to understand how AI products get built, prioritized, and adopted inside a large enterprise.
An SOP-writing AI tool, owned end to end. I designed and built an internal AI tool that automates standard-operating-procedure writing, turning a slow, manual documentation task into a fast and consistent one. I drove adoption to 100-plus users across Home Lending Finance in 2 months through targeted pitch presentations, cutting SOP documentation time by 75%.
What shipped, and what I cut. I shipped speech-to-text capture, Zoom-summary-to-SOP conversion, and in-tool feedback and bug reporting driven by live user input. I deferred email drafting and a general formatting chatbot to keep scope on the core use case.
Third of roughly 600 employees. I was product lead for Devvy Desktop, an AI work copilot, and we placed 3rd of ~600 employees in the Global Hackathon.
The design call that got us there. I kept the default UI minimal for non-technical users and put engineers behind an opt-in terminal mode, which is what let the product reach 80% of JPMC staff rather than only the technical ones.
Policy-to-control mapping, and I wrote the code. Alongside the internship, I helped an applied AI/ML team build a tool that lets regulatory controllers automate the manual assignment of policies to risks to controls. It was built in Fusion, a Python-based tool. Using vector embeddings and cosine similarity, it compresses a 2-day manual review into a 10-minute output check.
Headcount and budget forecasts. As a Global Finance and Business Management intern, I projected headcount and budget forecasts for call centers and operations teams, working in Excel with Dodeca and Essbase.