Where AI work meets operational discipline.

Keisy Caceres, Operations Lead. I run human data operations, AI training teams and workflow automation.

Get in touchView work
Portrait of Keisy Caceres

Where I've worked and what I build with.

Scale AIThe Exit SchoolReserve ProtocolArq (formerly DolarApp)MakeGoHighLevelLovableSlack

An industrial engineer who runs operations like a product.

Six-plus years in operations, five of them at Scale AI recruiting, training and managing experts for AI projects. Today I'm Chief of Staff at The Exit School, supporting two founders across a five-company portfolio and building automations in Make.com, Gemini AI and GoHighLevel.

Contributor operations

Recruiting pipelines, screening automation, onboarding and training, KPI and quality tracking.

Automation & systems

Make.com, Gemini AI, GoHighLevel, workflow design, SOP documentation.

Stakeholder management

Founder and executive support, cross-functional coordination, executive communications.

Tools

ClickUp, Google Workspace, Fathom, Intercom, Notion, Airtable, Slack.

Education: B.Sc. Industrial Engineering, UNEXPO (2012–2019). Certifications: Oracle Cloud Infrastructure AI Foundations Associate (2025), AI for Project Managers, LinkedIn Learning (2025), Applied Scrum, University System of Maryland (2022), Introduction to Data Science, IBM (2021), Data Science Fundamentals, Coderhouse (in progress). Languages: Spanish native, English advanced.

I turn messy, high-volume operations into systems that run. Pipelines that screen thousands of applicants. Models that cut cost. Automations that follow up on their own.

How I got here.

A career in order. The longest chapter was Scale AI.

Scale AI · San Francisco (remote) · Jul 2020 – Jul 2025

Five years, four roles, one pod of 100+ experts.

100+

experts across multiple countries in one pod

$5.2M

annual budget under my management

90%+

SLA performance, held

Bootcamp Lead Trainer

Jul 2020 – Jul 2021

Designed and ran one-week remote training programs for new contributors, with classes of up to 40 who moved into production at the end of the week. Every trainee passed and went on to deliver high-quality work.

Defined performance KPIs and updated the training based on quality results.

Team Lead — LATAM

Jul 2021 – Apr 2023

Supervised a 10+ person regional team, set its KPIs and built the remote onboarding and delivery process for the region.

Coding Pod Lead

Apr 2023 – Apr 2025

Managed ~70 Queue Managers and tracked the quality of the tasks their contributors delivered on AI coding projects.

Built a model that allocated staff between OPEX and COGS by project workload: OPEX −9%. Added validation steps agreed with Quality Control: PDR −7%.

Partnered with coding bootcamps in Buenos Aires, hiring about 10 people per bootcamp at a hire rate above 90%.

Workforces Pod Lead

Apr 2025 – Jul 2025

With the Data team, built an automated screening pipeline that filtered 2,000+ applicants and doubled pod headcount in 2 months.

Set role-specific KPIs with Trust & Safety, Enablement and Operations to track expert performance and work quality.

95%+customer satisfaction
~$2Min wallets secured

Reserve Protocol · San Francisco (remote) · Jul 2021 – Sep 2022

Support and liquidity in a crypto app

Product Operations Analyst & Liquidity Supervisor. Kept customer satisfaction above 95% on Intercom support.

Managed liquidity and real-time pricing for all trading pairs in the app, and applied security protocols to digital wallets (FTX and Trust Wallet) holding ~$2M.

5companies
2founders supported

The Exit School · Remote · Apr 2026 – Present

Chief of Staff across a five-company portfolio

Support the two founders on operations, internal communications, hiring and automation. Built an interview scorecard for hiring a CRM Coordinator.

Handle contractor onboarding, fix automation failures and write SOPs so the team can run recurring processes on their own.

Things I've built.

Four things from my work. The interfaces are recreated with fictional data.

Human data pipeline for AI training

Scale AIScale AI · Coding Pod Lead

Turns domain experts into high-quality training data, at scale and with quality control at multiple levels.

As Coding Pod Lead, I managed ~70 queue managers, each running the pipelines and contributors of their project, and tracked the quality of the tasks they delivered.

~70queue managers led
Thousandsof contributor hours
Leadership Me · Pod Lead ~70 queue managers Workforce lifecycleContributor DBOnboardingTraining & toolsProductionQueue managerQueue managerQueue manager…contributorsAttemptL1 ReviewL2 ReviewQuality CheckClient deliveryAttemptL1 ReviewL2 ReviewQuality CheckClient deliveryAttemptL1 ReviewL2 ReviewQuality CheckClient deliveryrework+ many more pipelinesClient hours

Swipe to see the full diagram

TaskRework: back to Attempt until it passes reviewActive contributor

Deal sourcing automation

Turns opportunities in GoHighLevel into data a web page can display, with no copy-pasting.

  • Make.com
  • GoHighLevel
  • Webhooks
  • HTTP

Off-market deals page

A live page of off-market deals fed from GoHighLevel. Every click on "I'm interested" alerts the team in Slack.

  • GoHighLevel
  • Make.com
  • Slack

Deal Evaluator

A calculator that scores acquisition opportunities on price, EBITDA, financing and DSCR in one view.

  • Lovable

Selected work.

Four systems I built to make operations scale.

From meeting transcript to client follow-up, hands-off.

Make.com pipelines that turn meeting transcripts into client follow-up emails, with a router that separates internal meetings, cohort calls and 1:1 coaching calls.

Fathom→Gemini AI→GoHighLevelGoogle DriveGmail
InternalCohort calls1:1 coaching

Staffing by actual workload.

A model that allocates staff between OPEX and COGS based on project workload.

−9%

OPEX

Automated screening.

A pipeline built with the Data team to filter applicants at scale.

2,000+

applicants filtered; headcount doubled in 2 months

Hiring developers straight from bootcamps.

Partnerships with coding bootcamps in Buenos Aires to recruit developers for AI coding projects, about 10 hires per bootcamp.

90%+

hire rate

A world, in development.

A hand-illustrated 2D puzzle-platformer built in Godot 4. Engineering with product discipline, in service of something I want to see exist.

Untitled project · Coming soon

EventBus

Systems talk through signals, not direct references.

Data-driven Resources

Parameters live in data, so tuning never touches code.

Light2D + normal maps

Illustrated surfaces that react to light.

GPUParticles2D

Atmosphere rendered on the GPU.

Layered parallax

Depth built layer by layer.

Results, in numbers.

6+

years in operations

2×

pod headcount in 2 months through the screening pipeline

2,000+

applicants filtered automatically

−9%

OPEX, by allocating staff by workload

−7%

Problem Delivery Rate, with new validation steps

95%+

customer satisfaction on Intercom support

Have a team or process that needs to scale?

Write to me. I reply in English or Spanish.

Get in touchLinkedIn

caceres.keisy@gmail.com