← All experience

Dec 2025 — Present · Bengaluru

Karbon.

Software Engineer II

Building AI-powered accounting products and full-stack systems. Working with product and engineering to bring complex features from an idea through to delivery.

  • Python
  • FastAPI
  • Amazon SQS
  • Temporal
  • LangSmith

My role

At Karbon, I work across the frontend, backend, and asynchronous workflows behind AI-powered accounting products. My scope spans building new modules, improving how large datasets move through the system, and bringing agent-driven products from architecture to implementation.

Accounting, from upload to sync

I built and launched the Accounts Receivable module across frontend and backend, covering AI-assisted invoice creation, allocation matching, and Tally synchronization. It reached 200+ uploads and 50+ active users in its first week.

For files with 40,000+ rows, I rebuilt invoice ingestion around streaming parsing and chunked asynchronous processing. A virtualized review grid with inline validation and bulk correction made those files practical to work with. Amazon SQS supported asynchronous processing and integrations with accounting systems.

Agent-driven products

For FlowTax, I owned the architecture and full-stack development of a CPA tax-filing platform, including client management and agent-driven, multi-step returns.

I also built Shadow PM, a multi-agent platform that connects Jira, customer transcripts, and codebase analysis for product research and UI prototyping. Python and FastAPI provide the backend, Temporal coordinates the workflows, and LangSmith makes their execution traceable.

Working with the team

The role also means helping teammates through difficult implementation details, improving the original brief, and coordinating with product to deliver complex features on time and with quality. I care about the full path from a requirement to something people can use reliably.

Let’s talk about the work ↗