I'm Kaushal Kumar — a VICIdial architect and cloud telephony engineer. I design predictive dialing platforms that stay up under load, keep the compliance side defensible, and scale without a rebuild. Founder of Adoptiv and VICIfast.
Working on dialers, carriers and call-centre infrastructure since 2008.
Installations, migrations and rescues delivered across managed hosting and direct engagements.
Concurrent agents on one deployment, with predictive dialing and live supervisor monitoring.
Daily call volume across platforms built and operated by Kaushal and his teams.
Cluster architectures that add dialers, carriers and agents without re-platforming.
US-based operations serving call centres across North America, EMEA and APAC.
Ventures
Everything I build serves teams whose business is the telephone — the dialer, the CRM behind it, the engineers who keep it running, and the campaigns that feed it.
Founder · Est. 2026
The self-writing CRM.
Predictive dialer, CRM and AI as one platform, built so reps talk instead of type — and so data stops falling through the seam between dialer and CRM.
Founder · Est. 2026
The fastest managed VICIdial. Live in 40 seconds.
Fully managed, cloud-hosted VICIdial on the unmodified open-source codebase — with root SSH kept intact — plus the pieces stock VICIdial never shipped.
End-to-end VICIdial engineering and support.
Engineers for teams running their own stack — hosting, carriers, DIDs and 24/7 support across every component of the call-centre platform.
Co-founder & Director · Est. 2024
Lead generation, software and AI delivery at scale.
The India-based engineering and campaign arm behind the group — custom software, AI integration and outbound lead generation.
Selected work
A sample of the problems people bring me — usually a platform that grew faster than its architecture.
Press-1 is unforgiving infrastructure: nearly all the work happens on calls that never reach a human. This one carries 5,000+ concurrent attempts, detects machines with its own AMD, and bills to six decimal places per tenant.
Cutover of the largest single floor I have run — 1,200 concurrent agents — from ageing on-premise hardware to a managed cluster, executed live between shifts.
Every patch meant a maintenance window, and every maintenance window meant lost dialing. Moving signalling off the dialers made nodes disposable and upgrades boring.
Writing
Things I end up explaining often enough to write down.
Managed VICIdial existed already. Almost all of it quietly forks the source, and once your build stops being VICIdial you cannot leave. That is the thing I wanted to fix.
Every calling operation runs a dialer joined to a CRM by an integration, and every one of them leaks data at that join. The fix is not a better integration.
The federal window is the easy part. What actually breaks compliance is that your dialer does not know where the person it is calling physically is — and the area code will not tell it.
Migrations, rescues, compliance problems, capacity planning. Tell me what's breaking and I'll tell you whether I'm the right person for it.