Skip to content
Andrei Csatlos

Case study 02

ATLAS Analytics

Software Engineer, then Technical Lead · Jan 2024 - Apr 2026

Singapore, remote · B2B sales and lead generation

Rebuilt a B2B sales platform in about two months, became tech lead within five, and built the AI agents the sales team used every day.

  • 50%+

    faster after the database redesign

  • $900 → $80

    per month, after replacing Smartlead

  • Weekly → daily

    release cadence

The situation

SalesOS, ATLAS's B2B sales and outreach platform, had outgrown its first data model: new features got harder to add, the same bugs kept coming back and running costs were higher than they needed to be. I joined as a software engineer and was put in the lead role after about five months, working directly with the founder several times a week.

What I did

  • Rebuilt the core platform with one other engineer in about two months, then took over its architecture and delivery across frontend, backend and data.
  • Redesigned the database properly (normalised, indexed, cleaned of orphaned and test data). The product got more than 50% faster, recurring bugs dropped from about 10 a week to about one every week or two, storage went from 70% to 25% and infrastructure costs roughly halved.
  • Designed a hybrid EAV data model so customers can add their own fields to records, Salesforce-style, without slowing queries down.
  • Built an in-house CRM on EmailEngine for scheduling, balancing and sending email. It replaced Smartlead and cut that cost from about $900 to $80 a month.
  • Built AI agents the team relied on: one sorts about 200 email replies a day, answers prospects and enriches the database; others write personalised email, WhatsApp and LinkedIn sequences, watch prospects for buying signals, and turn meeting transcripts into a knowledge base you can search.
  • Let management build reports and dashboards by chatting with the database, with access rules enforced in code rather than left to the AI.
  • Moved the team from weekly to daily releases by bringing AI into testing, security checks, code review and business-impact checks.
  • Mentored 6 engineers (3 junior, 3 senior). One grew into a Product Owner role and took over a product, which freed me to focus on the others.

Stack

  • TypeScript
  • Node.js
  • React
  • SQL data modelling
  • AWS
  • EmailEngine
  • n8n
  • LLM APIs