AI assistants & document processing
Build assistants that answer from your information, extract data from documents or turn recordings into useful summaries.
I’m Arushi Prasad, a senior software engineer with experience at Moveworks and Goldman Sachs. I build AI products, internal tools and dependable backend systems.

Build assistants that answer from your information, extract data from documents or turn recordings into useful summaries.
Replace repetitive admin with clear tools for approvals, reporting, bookings and day-to-day operations.
Design and build backend services, integrations and analytics pipelines, with attention to reliability and access control.
A closer look at the problems, the software and how it works. Client identities are kept private; screenshots use sample data.
Staff, budgets and spending approvals in one system, with permissions tailored to each role.
A confidential booking flow that matches students with volunteer counsellors on WhatsApp.
Daily sales and cost reporting built around WhatsApp voice notes, with kitchen-level dashboards.
Extract printed text, tables and handwritten notes from scanned invoices.
Designed analytics query services and secure reporting features. Benchmarked ClickHouse with Kafka ingestion and production-scale workloads; implemented Snowflake data retention and designed permission checks for data connectors.
Led 5–7 engineers delivering employee-profile and mobility features. Built data pipelines, an AWS email framework and event-driven services, spanning Kotlin backends and React interfaces.
Built Spark pipelines processing 100 million records to power relationship discovery and search. Developed interactive network visualisations and an API performance-testing framework.
Earlier: Goldman Sachs engineering intern, Jan–May 2019.
Questions by text or voice, answered in English and Hindi from an approved charity FAQ.
Unlock a password-protected PDF using a password you already know. Files stay in memory.
Search any city for current conditions and a 24-hour forecast.
These demos replay sample workflows so you can explore the output without running paid AI services.
Turn English, Hindi or Hinglish recordings into speaker-labelled transcripts and clear next steps.
A replay of turning a public Facebook video into a transcript and structured study notes.
Explore the research workflow through a replay of its stages and sample output.
Tell me what you’re trying to improve, who will use it and what a good result looks like. A rough idea is a fine starting point.