I build the operating systems behind post-sales teams.

Problem

The company had brought in Dust, a no-code agent platform. Agents for CSM meeting prep already existed and worked, built by a colleague on the technical side. Nobody used them.

The pilot was drifting, and the SVP was not going to approve continued access for CS without evidence the thing earned its keep. No evidence meant CS loses the platform.

What I built

The agents were someone else's work. Mine was everything that turned them into something a team actually uses, and getting people to use them turned out to be the whole job.

I sat with the CSMs and explained what the agents were for and what was riding on it: no proof of value, no platform. I opened a Slack channel so the pilot lived in one place, answered questions as they came up, and when somebody went quiet I asked them why they had stopped rather than chasing them.

6 CSMs on live accounts, no barriers in their way. Then I collected a time-saved estimate from each of them and turned it into a number the SVP could act on.

Result

Time given back to the team, self-reported by the 6 CSMs who used it:

5.5 hours a week

That number won the approval. All 6 are still using the agents, and the rollout to the rest of the team goes out as they are, with nothing further to fix.

Problem

Net revenue retention forecasting was inconsistent across regions. Every leader had their own definition of what counted, so the roll-up never quite reconciled and the number that reached the exec team came with an asterisk.

What I built

A forecasting framework with one shared definition, one input process across regions, and Salesforce reporting behind it. I own the renewal quote and approval flow that feeds it.

Result

One NRR number that leadership works off, instead of three that disagree.

Problem

Cloudinary bills on usage, and customers went past their limits regularly. The platform did warn them automatically, but plenty of customers never registered the warning.

That left two bad endings: invoices that were hard to collect, and customers hit with bill shock. Neither is a place you want to be with an account you are about to renew.

What I built

The whole thing, end to end. The data team surfaced the overages. Everything downstream of that signal was mine: a notification layer that routed each case into the right channel in front of the right people, sized by two variables, how big the overage was and how big the account was.

On top of it, a rule set for who reaches out, to whom, and how, across CS, Finance and Sales. I owned both halves, the technical plumbing and the agreement about who works with whom.

Result

Visibility first. Overages stopped being something the company discovered at invoice time. Then the part that mattered more, out of every overage case we knew about:

100%

ended with a person contacting the customer, rather than an automated notice we hoped they had read.

Problem

A post-sales organization of 90+ people across multiple regions, running on processes that had grown organically rather than been designed.

What I built

I built the CS enablement function from scratch and owned the operating model, KPIs and systems for a CS Operations function serving 50+ post-sales people across regions.

Salesforce at admin level: objects, fields, automation, reports, dashboards, and data integrity.

I also ran the replacement of the core CS systems stack end to end: vendor evaluation, migration, and the internal launch.

Result

One operating model and one set of KPIs across every region.

After the systems stack replacement, 70% of users reported higher productivity and 90% reported satisfaction with the new stack.

Problem

Thousands of support tickets a month, handled by an internal team, on a Zendesk instance that needed to route and categorize them correctly without a human triaging every one.

What I built

I configured Zendesk rather than just using it. I owned the tagging taxonomy, the macros, and the routing and automation rules.

Same hands-on depth in Intercom, and in Jira and Confluence: custom workflows, boards, fields, and Confluence spaces I built myself.

Result

A support operation that scaled on configuration instead of headcount.