Product Engineer · Ex-Founder
Aarav
Khanduja
Product Engineer. Spent the past year at Phia, building internal products end to end — data pipelines, analytics infrastructure, and applied AI.
Experience
Across startups, internships, and side projects, I've focused on building products people actually want — and validating that with data and usage.
Product Engineering Intern
Agentic Systems & Evals
- Shipped a browser-agent validation pipeline (browser-use on Kernel) that cut false positives 3×, selected through a controlled two-model bakeoff and used to automate QA.
- Built an AI eval lab that runs 300+ real support-ticket prompts through the production pipeline in isolation, letting the team benchmark and iterate on features without deploys.
Data & Analytics Infrastructure
- Re-architected analytics with nightly materialization, replacing multi-GB warehouse scans with millisecond reads and enabling faster performance analysis.
- Built an end-to-end user segmentation system across the data pipeline, targeting engine, and UI, enabling personalized CMS targeting that increased engagement.
Content Systems
- Built an image enhancement review and publishing pipeline with staged review, CDN delivery, and bulk operations, scaling enhanced home-feed assets and increasing CTR and session time.
Product Management Intern
- Partnered with ML and engineering to turn messy user feedback on search relevance into ranking improvements, reducing latency by 80% and increasing product clicks by 45%.
- Built an analytics MCP server and unified OAuth gateway (1k+ LOC, 18 commits), letting LLM clients query product analytics — still actively used across the product team.
- Built a structured research system, interviewing 70+ users across active, churned, and new cohorts, driving 10+ shipped experiments and contributing to 100K+ retargeted clicks.
Joined pre-seed at under 10 people; the product grew to 1M+ users and a $35M Series A led by Notable Capital, with Khosla Ventures and Kleiner Perkins.
Product Management Intern
Doubled engagement on the core “Learn Mode” feature by reducing time-to-activation through onboarding UX, CTAs, upload performance, and AI-generated content.
Used Mixpanel and session replays to analyze onboarding funnels and inform product design.
Cofounder
ProperlyAI
- Partnered with 5 real estate teams across Compass, Coldwell Banker, and Douglas Elliman to validate workflows and product direction.
- Placed 2nd at Techstars Startup Weekend and joined the NYU Sprint Accelerator, securing $6k in grant funding.
- Ran 70+ user interviews and onboarded 5+ beta teams, driving 3 major product pivots.
- Owned product, GTM (cold outreach), and execution across a 5-person team.
- Built and shipped an AI CRM for realtors, turning inbox data into structured client summaries.
Product Marketing Intern
Integrated Mixpanel into the frontend and designed a UTM strategy analyzing 10K+ visits across 1M+ impressions.
Helped drive early adoption of an AI lecture notes → quizzes product across student communities in Harvard, Columbia, NYU, UCLA, and UCSD.
Built a gamified rewards system with a leaderboard that awarded points for event participation and enabled members to redeem points for merchandise, increasing engagement across a 40+ member community.
Increased event attendance through incentive-driven engagement loops.
Featured Projects
Open-source projects I've contributed to.
Built local-first LLM support for a context-aware text editor, removing dependency on cloud APIs and enabling offline usage.
Contribution
Impact
Tech
Improved editing workflows for a cursor-free screen recorder focused on power users.