I'm Megha. I spent four years helping build Looppanel, a Sequoia-backed AI research platform, from its first designs to an enterprise AI product: first as founding designer, then as AI Product Manager. Now I'm at Northwestern, doing an MBA at Kellogg and an MS in Design Innovation through the MMM program. I also build and ship my own products with AI coding tools.
The business problem, the bet, the trade-offs and the results.The research, the explorations and the craft behind each decision.
Win/loss analysis showed we were losing on how our AI felt, not what it could do. I made the case to collapse the workflow into one AI experience, scoped v1 and launched it. The market’s verdict flipped from “not at par” to “the best AI in the market.”I wrote the prompts before opening Figma, tested every competitor's AI hands-on and QA'd each staging build. The result was a wizard that turns interview transcripts into grounded summaries with follow-up questions.
An email-based document step was causing drop-offs in a 5–7 day lending funnel and tying it to ops headcount. I made it self-serve in a one-month sprint: 100% in-app, with no dependency on support.Turned an offline, friction-heavy document submission process into a guided in-app experience, with layout explorations, reusable upload components and error states.



Market data, teardowns and interviews traced Snapdeal's bounce rate to a trust problem, and led to a strategy built on local tailors and shops. Published at ICoRD’21 (Springer).A heuristic evaluation, 10 interviews, two personas and a prototype covering tailor-verified fit, photos of the exact item being shipped and neighbourhood stores.
Designed for both the relief agency and the person eating it: a ₹25–45 pack that grows microgreens while stored. Patent applications filed in India and the US.Visual research, recipe testing, seed selection and flood-proof packaging, with a collapsible seed compartment inspired by origami.
A running log of products I've taken from idea to working software with Claude Code and Cursor.

Chrome extension that adds cycle phases to Google Calendar.

A paid vocabulary product that drills the words you get wrong until you know them.

A local-first AI voice companion that speaks up before your calendar events.

An interactive history of AI from 2018 to 2026, with a practical note on what each milestone can actually do.

Offline-first wayfinding for Kellogg’s Global Hub, written like a friend’s directions.

A Chrome extension that strips images and video from Instagram so there’s nothing to doomscroll.
I started as a designer, and I still work like one: I start with real users, test ideas hands-on and build things to learn. What has changed is how much I own. Today that includes the strategy, the roadmap and the results.
I joined Looppanel as its founding and only designer, and defined the core product from transcription and note-taking to analysis and the research repository. I then built its monetization and growth features. As AI Product Manager, I led the pivot to our AI Insight Wizard, owned prompt engineering and model evaluation, and steered the move to enterprise customers, which lifted annual revenue 270%.
I prototype in code. On my team, Claude Code and Cursor replaced static design handoffs with production-ready front-end code. I've also built and launched my own products, including a menstrual-cycle productivity tool used in more than 75 countries.
Before Looppanel, I was a product designer at Razorpay, India's largest payment gateway, where I started its user research program and designed lending products. I hold a B.Des from IIT Guwahati, where my student work led to a published paper with Springer and patent applications in the US and India.