Turn a recurring manual process into a system.
A team describes the same weekly pain three different ways. I ask the questions that find the real workflow, whiteboard the fix, and assess if an automation has positive ROI.
Automation Systems — Mexico / US Hours
Guanajuato, MX
Víctor Mahatma
I build the systems that move your business forward.
Internal tools, automations, and AI agents for teams that have outgrown spreadsheets and manual handoffs. I talk to the people doing the work, ship a working version in weeks not months, and document it so the next person can pick it up.
Tools
Tools, Frameworks and what I’m testing next.
- In use
- Testing
Selected systems / 01–04
Four systems. Each one replaced recurring manual tasks, saved time and money.
01
02
03
04
Voice agents that make the call and book the meeting
Autonomous voice sales agents
Built on
ElevenLabs + Twilio
→
Lead lists that surface businesses with no website
LocalLeads · Desktop app
Built on
Google Places API
→
Onboarding that starts the moment the contract is signed
Client onboarding · Automation
Built on
ClickUp + Slack + CRM
→
One master video, localized for every market
Multi-language video localization pipeline
Built on
Gemini + Descript + HeyGen
→
Work Philosophy
I treat every automation as a small product with a lifecycle, not a script with a demo. I’m less interested in what an agent can do once than in what it will do the hundredth time, when the data is half-missing and the person relying on it has no patience left for error messages.
Connect the tools that were never meant to talk.
I can build custom tools when existing connectors are not enough; n8n or Zapier when speed is priority.
Build reliable systems that a team can depend on.
Structured outputs, tool calling, model selection, cost control, test cases, scheduled reviews, and failure alerts. An agent that works in a demo is not the same as an agent that works on a Tuesday.
Tools a non-engineer can run.
Friendly documentation, guides, and a written record of every architecture decision — including risk factors. The system should survive my absence.
A system that works even when things break.
A fallback layer: retries and queues for flaky integrations, safe defaults when data is missing, duplicate-event handling, back-end validation, and a manual override when automation can’t be 100% trusted.
About
I am a full-stack developer and automation builder based in Guanajuato, Mexico working with distributed teams on their internal systems. I write JavaScript and Python, tune LLM agents, and build the small internal apps that hold a business together. I have been writing code since before AI made it easy, which means I can read what the models produce, find what is wrong with it, and fix it before it reaches production.
C1 English or native Spanish. I am direct about what is worth building. If the manual process is cheaper than the system, I will say so.
Time in Guanajuato
Today
Timezone
Based in Mexico · US Central Time · Bilingual English–Spanish · Available for remote work
Prompt library / 9 prompts
The prompts you can use to scope, assess and evaluate automations. Copy them and swap in your own context.
How to use
Every prompt follows the same protocol:
- Frame
- Extract facts, assumptions and unknowns
- Infer constraints and failure modes
- Generate options
- Stress-test against scale, adoption and maintainability
- Decide with tradeoffs
- Produce a handoff artifact
Contact
Tell me what your team does manually every week.
I will reply with an honest read on whether it is worth automating, what it would take, and what it would cost. No discovery-call theater.