This past summer, I interned at GitHub as a Product Management intern within GitHub Advanced Security (GHAS), their security products org

• They recently announced LLM-based Code Security products that work on top of CodeQL, their existing static analysis engine
• I took ownership of the evaluation strategy, focusing on accurately evaluating how well our models find vulnerabilities in code

This is easier said than done. LLMs have many common pitfalls when it comes to finding vulnerabilities within code

• Issues like training contamination or false positives challenge the accuracy of evaluations, and we needed to make sure we took this into account during evals
• Building the evaluation framework meant running LLM-based assessments across an internal corpus of real code, benchmarking detection and remediation quality against existing tooling like CodeQL
• Many companies are investing in this space, and these competitors were putting out marketing showing shallow evaluations with questionable accuracy

We needed a base evaluation strategy define both our internal narrative (metrics to shoot for as we build and refine this tool), and external (how to we communicate our tool is the best to customers)

• I quickly got up to speed via conversations with engineers and AI/security researchers already deep in the product
• I met with many enterprise AppSec teams to understand how security-conscious organizations actually evaluate and trust AI-driven tooling before they let it near production code
• I researched what competitors were putting out to get an understanding of their narrative, and how GitHub can differentiate
• I also researched as much data as possible from internal data warehousing + our evaluations of models across our corpus to pressure test and support our thesis on this

By the end of the summer, that work turned into an external-facing security blog post. Something we can share with customers that detail how GitHub is on top of the vulnerability scanning space

I co-founded a parking management platform company called Parkd

• It is an end-to-end system delivering permit registration, vehicle management, and remote parking enforcement to schools and universities
• We use a trained license plate reading ML model and adapt it to existing school cameras to remove the need for physical parking inspection
• We have both an admin-side for managing and enforcing parking, and a user side, for buying and managing permits

We're partnered with Santa Clara University (our school) and Bellarmine College Prep (a local high school) for pilots of our platform

• We began outreach around June of 2025, a few weeks after we began working on it, and rolled out to Bellarmine in September 2025
• Their entire parking management system runs on Parkd, managing an estimated $100K per year in school revenue via parking permits
• With SCU, we're still integrating with our school's camera system, but progress is coming soon (lots of bureaucracy and approvals to overcome first)

I worked as a Product Management intern for Permiso, a leading and fast-growing identity security startup

• Collaborated with cross-functional engineering, UI/UX, and GTM teams to align customer needs with company OKRs
• Wrote 100+ user stories/Jira tickets directly tied to code execution, reducing backlog & enabling faster feature delivery
• Designed and validated a per-user pricing model for SaaS and cloud infrastructure integrations by running SQL queries on AWS, resulting in improved sales pipeline conversion, influenced $2M in potential ARR.
• Developed 'trust score' framework to quantify identity risk posture by combining identity factors with ML-based dynamic behavioral analytics, improving how F500 customers identify high-risk identities.

Worked as a sourcing and diligence extern for one of the world's leading venture capital firms

• Highly competitive externship, one of two non-targets on Dallin Bill's team
• Scoured the internet for B2B SaaS startups that fit our team's thesis, pitching 5 startups weekly onto our CRM. Battery analysts would review the startups and select those to move forward with a founder call
• Drove the diligence process for 3 startups I sourced, leading founder calls, going deep into their vertical, mapping competitive landscape, and making a harvey-ball analysis of their product
• Ultimately, I presented these to general partner Michael Brown

Worked as a "Private Equity Research Intern" (essentially an SDR) for a private equity advisory firm

• I sourced acquisition targets for our clients, did preliminary research on Pitchbook, Sourcescrub, and LinkedIn Sales Navigator
• Logged the research onto our Salesforce CRM to be reviewed by full-time analysts
• Also sent 100s of cold emails, and made cold phone calls pitching our client's financial services to target clients

Began working for Ergoworks over the summer in high school at age 16, a startup selling ergonomic office supplies on Amazon and Shopify

• I managed our back-end operations, printing out daily orders, boxing and packaging them in the warehouse
• Managed our AWS database of products, updating price changes, logging new products, and removing old ones, updated website in HTML
• Became proficient in Microsoft Excel, AWS, and gained valuable experience in attention to detail and professional workplace dynamics