Navya Shah
Product & Ops · USC '26 Building Local, currently at Spin.AI 📍 Los Angeles, open to relocatecoming soon! 🥘
I find things that are broken and fix them because I can't not notice.
I've wanted to build something of my own ever since I can remember. With technology growing at unimaginable speeds, this could be my own app, startup, or even my personal brand.
I started off on the traditional finance path, study economics and then become a financial analyst/investment banker. But real satisfaction came when I challenged myself to learn something I always doubted myself a lot to pick up: coding.
Taking on a Data Science double major let me explore learning that I truly enjoyed. It took me from learn the basics of python to being a teaching assistant for OOP for about 2 years. I learned how to communicate the technical to non-technical, developed problem solving in myself and others, all while building a community along the way.
I stayed on at USC to finish my Master's in Cybersecurity Engineering through my DS degree. I know, big jump but it was well worth it as I conquered understanding computer systems on an infrastructure level all while learning best security practices.
My unique background brought me to my perfect fit: product management. Really a role where you get to play entreprenuer simulator and own everything you research and build. I truly enjoyed behind the scenes and making the features users actually touched. I started by doing this back home in Mumbai at Quantech Datascience for a tax compliance app pre-launch and continued it over in LA where I worked with Spin.AI on their cybersecurity platform.
The thread through all of it has really been: operating and building with real stakes attached.
Right now, I'm in love with building/designing software with Claude Code and Codex. This is curently what I'm focused on, shipping apps I like while also documenting the build, what it looks like to ship a consumer AI product as a solo builder.
If any of this resonated with you...
- Analyzed 20+ SaaS products to identify feature gaps and built 100+ security control mappings.
- Defined requirements for a Value-Added Partner security controls feature: assessed how partner configurations would interact with platform-level settings, mapped edge cases, and wrote a BRD.
- Identified that solutions engineers had to fully log out and back in to switch client accounts; designed the UX flow to resolve it and validated the pain point with customer success before scoping.
- Collaborated with engineering on acceptance criteria, edge case identification, and sandbox integration.
- Evaluated security controls across the database, API layer, file storage, and user management for Ritetax, a tax compliance product handling sensitive financial data for chartered accountant firms.
- Identified that client data separation was enforced at the UI layer. Tested to confirm unauthorized data access, documented the gap reproducibly, and escalated to engineering for a pre-launch fix.
- Detected through log analysis that the OCR pipeline silently returned incomplete data. Recommended a post-extraction completeness check that now flags bad scans immediately.
- Built SQL-driven loan default tracker analyzing 1,000+ accounts, improving accuracy of risk monitoring by 20%. Done by using behavioral buckets to identify risk patterns after examining disbursement data.
- Synthesized market data into insights that guided portfolio strategy and risk modeling discussions.
- Supported 100+ students through help sessions and assignment reviews, clarifying coding concepts.
- Debugged student code in Java OOP assignments, identifying logic errors and edge cases in real time.
- Led problem definition for an Air Force contracting visibility system: synthesized insights across stakeholder interviews, and defined the system architecture for an LLM-powered Text2SQL pipeline that enables non-technical users to query fragmented contract data in natural language.
- Building a RAG pipeline using pgvector and Supabase to surface hyper-local restaurant recommendations from unstructured data across reviews, ratings, and creator content. React PWA frontend with intent classification and conversation layer powered by Claude.
Here's what I've shipped and worked on.
Email is the best way to reach me at navyashah6@gmail.com