01Consulting02Founder scorecard03Story04Operating record05Professional profile06Builds07Writing08Creative
LinkedIn ↗GitHub ↗Email ↗
Evan D'Souza
LocationBengaluru, India
BuildingDszape · BeckyOS
Background10+ Years Startup Ops
Currently IntoAI Agents · Local LLMs

The Decade in Operations

I spent ten years as the first operations hire, the person who turns chaos into systems. At WeavedIn I scaled a POS SaaS from 4 accounts to 2,500before Paytm acquired it. At Paytm I engineered ₹96 crore in annual revenueand launched India's first online deployment of payment devices. At UrbanPiper I ran50+ simultaneous B2B onboarding projects; at Aviyel I built a2,000+ member developer community from zero.

Then AI Changed the Maths

I studied computer science but never worked as an engineer. In 2026 that stopped mattering. I built Dszape, a full hotel management platform ("Shopify for hotels"), completely solo, to learn what production-grade really takes: booking engine, front desk, folio billing, GST invoicing, and a CMS website builder. 245+ API routes, 180+ database migrations. AI writes most of the code; ten years of operations discipline keeps it correct.

Building solo forced me to solve AI's management problem, so I builtBeckyOS, a multi-agent coding OS published on npm. Specialist agents with separation of duties, evidence tiers instead of self-reported "done", and rules compiled from real production incidents. The agent that builds can't grade its own work.

The Lab: Local LLMs

Alongside the products, I experiment with what runs on my own machine: local models via Ollama, a local text-to-speech voice for my always-on assistant, file-based long-term memory for agents, an offline AI vocal mastering chain for my music label, and an MCP server that lets agents write MIDI straight into FL Studio. Local models where privacy and iteration speed matter; frontier models where capability does.

And Still Consulting

I keep three consulting seats for early-stage startups. It is the same hands-on systems work I've always done (ops, funnels, onboarding), now with AI adoption added. If you want the playbooks without the engagement, they're all on the blog.

Key Highlights

Dszape: Hotel SaaS, Built Solo

A full hotel management platform: booking engine, front desk, folio billing, CMS website builder. 245+ API routes and 180+ migrations, shipped by one person with AI agents.

BeckyOS: Multi-Agent Coding OS

Published on npm. Specialist agents with separation of duties, DONE/VERIFIED/AUDITED evidence tiers, and rules compiled from real production incidents.

4 → 2,500 Accounts at WeavedIn

Built onboarding, helpdesk, and training from scratch for a POS SaaS, scaled it 600x before the Paytm acquisition, then engineered ₹96 Cr revenue at Paytm.

50+ B2B SaaS Launches at UrbanPiper

Ran 50+ simultaneous onboarding projects across web, Android, and iOS for a leading Indian restaurant-tech platform, including compliance, stakeholders, and training.

Community From Zero at Aviyel

As Head of Global Operations, built a 2,000+ member developer community and onboarded 200+ technical writers and 1,500+ open-source contributors.

The Local LLM Lab

Ollama on Apple Silicon, a local TTS voice for a persistent assistant, agent memory systems, offline AI vocal mastering, and an FL Studio MCP server.

Let's Talk

Building something with AI, or want hands-on help scaling? I answer every serious email.