Melove Gupta

/miː lʌv ˈɡuːptə/•noun•

00:00:00IST

I'm a {Software} and a {Full-Stack} Developer...

I'm also a CS undergrad student and I spend most of my free time inside codebases that I didn't write. Over the past year that's meant fixing a floating-point bug in LibreOffice and building an instrumentation watcher for CNCF's OpenTelemetry organization...

Experience

CNCF OpenTelemetry Ecosystem Explorerlink

Open Source ContributorMay 2026 - Present

Built a new monitoring tool for JavaScript projects from scratch, modeled on an existing tool built for Java, that automatically tracks 46 entries in the project's registry, and went through 16 rounds of maintainer review before it was merged.
Fixed a bug in the Faro monitoring SDK that was sending local development data to the live production server by mistake, causing browser security (CORS) errors across the entire application.
Found and fixed a display bug that broke the layout on mobile phones, tracing it back to a webpage grid that wasn't sized properly for smaller screens.
Wrote onboarding documentation to help future contributors ramp up faster, and merged 22+ pull requests overall while collaborating directly with maintainers from Grafana Labs.

Other Open Source Work

LibreOfficelink

Open Source Contributor

Fixed a 12-year-old bug (tdf#55960) in LibreOffice Calc where auto-filling a column with a repeating number pattern produced tiny rounding errors instead of clean values (e.g., 0.6999999999999996 instead of 0.7), traced the issue to how the spreadsheet engine computed each value, corrected the rounding logic, and added an automated test to catch future regressions, merged after seven rounds of review and later adopted into a downstream release.
Diagnosed and fixed bugs that affected multiple LibreOffice applications at once (Writer, Calc, and Impress) by tracing how a single change rippled through the rendering, interface, and document-handling layers.

In Between These Experiences

Projects

Chain Lens - Bitcoin Transaction & Block Visualizer link

Built a Bitcoin transaction parser from scratch that reads raw blockchain data byte-by-byte with no external libraries, supporting all major transaction formats, modern ones like SegWit and Taproot, plus older legacy formats.
Verified block data hadn't been tampered with by rebuilding its cryptographic hash structure and checking it against the block header; made the parser usable three ways, a command-line tool, a REST API, and a web-based visualizer.

Pramaan - Citation-Grounded RAG System for Indian Law link

Built a search system that combines traditional keyword search with AI-based semantic search, then merges and re-ranks the results for accuracy, ensuring every AI-generated answer is backed by a verifiable citation from a 20-document set of Indian law.
Designed a 122-question test suite to benchmark 4 different AI models on accuracy and reliability, wired into an automated pipeline that blocks any change from being merged if it makes answers less accurate.

Safe PSBT Transaction Builder link

Built a Bitcoin transaction builder, following the standard PSBT format, that supports fee-bumping, scheduled transactions, and smart coin selection; wrote 15+ tests covering the full transaction lifecycle from construction to signing.

Card Drawer - Web3 dApp link

Built a full-stack blockchain app (Next.js/TypeScript front end, Solidity smart contract on a test network) that connects to a user's crypto wallet and keeps the interface responsive while transactions confirm on-chain.

Work in Progress

Parent Copilot for Autism

Professional autism support is expensive, scarce, and confined to sessions. But the situations parents actually need help with are things like a meltdown before school, a routine that suddenly stops working, or a new environment the child can’t process. The parent is often the most important person in the whole support system, and the one with the least continuous support.

Most tools in this space treat behavior tracking, visual schedules, and parent guidance as three separate apps. I think there should be 3 ways of interacting with one longitudinal record of a child.

I'm working on a loop where a behavior entry gives the Copilot context, the Copilot surfaces a pattern, and the parent adjusts a routine based on that.