A marketplace where working professionals sell their experience to people who need it, through 1:1 sessions, group classes, live courses and direct Q&A. Clients book and pay in the app, sessions run on in-app video, and operators are paid out through Stripe.

What was Abhishek Thakur's role on Uplevelit?

Development team lead at Bizdesire. He owned architecture and delivery with the team, reviewed code, and worked hands-on on the booking, payment, refund and payout paths.

What stack did Uplevelit use?

React on Firebase Hosting, Firebase Functions for the backend, Stripe for payments and payouts, Twilio Video for live sessions, Bunny Stream for recorded video, SendGrid for email and Looker Studio for reporting.

Uplevelit case study | Abhishek Thakur

The job

Uplevelit sells experience. A professional who has done the job offers what they know to someone standing where they once stood, as a 1:1 call, a group class, a live course or a direct question and answer.

I led the engineering team that built it at Bizdesire. Alongside architecture, reviews and delivery, I worked hands-on on the parts the business runs on: booking, payments, refunds and operator payouts.

Session formats and integrations (conceptual)

Four formats, one booking

A 1:1 call, a group class, a live course and an asynchronous question look like four products. Underneath, each one needs the same things: something to buy, a time or a deadline, a payment, access for the right people, reminders, and money owed to an operator afterwards.

We treated them as variants of one session model instead of four features. Booking, payment and notification code stayed singular, and a new format meant new rules on the session rather than another copy of the plumbing.

Where the money moves

A booking is not finished when the card is charged. The client needs access to the session, the operator needs the earnings recorded, and a cancellation has to reverse the right things.

Stripe handled purchases, refunds and payouts to operators, with webhooks updating the application after each payment event. Webhooks and client retries can fire twice, so the functions that touch Stripe were written to be safe to run twice: a repeated event must never charge or pay out again.

Sessions inside the product

Live sessions ran on Twilio Video inside the app, so neither side had to swap meeting links. Twilio handled the real-time media and the network traversal, and we spent our time on the experience around the call.

Recorded sessions, course videos and operator intros went to Bunny Stream, which covers transcoding, adaptive streaming and delivery in one service. Live calls and stored video have different jobs, so they used different tools.

Running without servers to babysit

The backend ran on Firebase Functions behind Firebase Hosting. For a team whose job was product, not infrastructure, that removed containers, scaling settings and a separate backend deploy. Each function owned one job, and the hand-offs between them were explicit.

SendGrid sent the transactional email a marketplace lives on: booking confirmations, reminders, receipts and account flows. Looker Studio sat on top of the data, so the business could read operator earnings and session numbers without asking the team for a report.

Leading the team

As lead, I worked with the client on scope and estimates, coordinated web and mobile delivery, reviewed code, mentored developers and ran technical interviews. I kept hands-on ownership of the booking, payment and payout paths.