WELCOME.txt
Synaptech Solutions is the independent engineering practice of Germán Martínez, based in Uruguay. I build production-grade AI, LLM and backend systems (AI assistants, ERP and MCP integrations, data pipelines) for teams in Latin America, the United States and Europe.
- whoami
- Germán Martínez
- Semi-Senior ML Engineer · Bachelor's in Software Engineering
- LLM Application Specialist
- Independent AI Systems Consultant
- cat mission.txt
- Practical software. Honest AI.
- Production-grade systems for teams that need AI to work reliably.
- ls expertise/
- backend systems
- LLM infrastructure
- agent orchestration
- data pipelines
- workflow automation
- applied AI integrations
- cat working-style.txt
- direct technical product discussion
- clear architecture before implementation
- async-friendly collaboration
- no vague AI hype
- honest tradeoffs, documented decisions
- uptime
- 7+ years building software systems
- independent consultant
- production-first mindset
- cat location.txt
- Based in Uruguay
- Remote-first
- Available for LATAM, US, and EU-aligned teams
bot.platform.architecture
§ 01 // method
HOW_I_WORK.md
Every engagement begins with a written technical spec, and I don't write code without one. That keeps scope from drifting, and we both know what "done" means before a line is written.
I work in documented, asynchronous cycles: you get a written update every week instead of daily standups. Your time is valuable, and the best way I can respect it is to be precise and not need supervision.
If your timeline is impossible, I'll tell you upfront. If your budget doesn't match the scope, we'll renegotiate before starting, so the invoice holds no surprises.
You own everything that's built: clean repos, documented APIs and runbooks. The point is that you don't depend on me.
WHO_I_DONT_WORK_WITH.md
If "we'll figure it out as we go" is your engineering philosophy, we're not a match. Ambiguity is expensive and I refuse to bill you for confusion.
I won't build LLM wrappers disguised as products. If you need "ChatGPT but for X" without a clear moat or defensibility thesis, I'll decline politely.
Pre-seed with a real prototype: possible. Series A company asking for "sweat equity": no. My rent is not a venture bet on your cap table.
User tracking without consent, engagement manipulation, or deceptive UX: the answer there is no. I review use cases before signing any agreement.
§ 02 // work log
click to expand§ 05 // manifesto
engineering philosophy"Software is not a product.
It is a decision
made permanent in code."
Every architectural choice, every abstraction and every shortcut compounds. That's why I treat code as a long-term liability rather than something you deliver and forget.
Complexity is a debt, not an asset.
Every layer of abstraction you add must justify its existence with measurable value. Teams collapse under architectures that impressed in a pitch deck but couldn't survive a Tuesday morning incident.
AI is infrastructure, not magic.
LLMs are probabilistic systems: they fail silently, drift over time and hallucinate under pressure. Treat them like deterministic APIs and you've shipped a liability. That's why I build with evals, guardrails and fallbacks.
The best system is the one that ships.
Chasing perfection usually delays the launch. I make the system correct first, then observable, then easy to maintain. A working system with known tradeoffs beats an elegant one that never deploys.
Honest engineering over comfortable lies.
If your idea is technically unsound, I'll tell you. I flag timeline risks before they turn into crises, and I write down the decisions I disagreed with. Being candid is part of the job.
Ownership means leaving, not staying.
My goal is for you to stop needing me. Every engagement ends with documentation, runbooks, and a team that can operate without me. Creating dependency is a common habit in consulting, and I avoid it on purpose.
- →Written specs before any code is written
- →Eval suites shipped alongside every LLM feature
- →Architecture Decision Records for every major call
- →Honest scope and timeline pushback when needed
- →No vendor lock-in without documented justification
- →On-call runbooks before any system goes live
§ 03 // technical stack
§ 04 // principles
Boring tech is a feature.
PostgreSQL before a distributed database. A cron job before a message queue. The most maintainable system is the one your next engineer can debug at 2am without Slack.
Evals before deployment.
Every LLM system ships with a benchmark suite. If you can't measure it, you can't improve it, let alone sell it to an enterprise customer with a legal team.
Runbooks, not tribal knowledge.
Every system I build ships with operational documentation: on-call runbooks, architecture decision records and deployment guides. Your team shouldn't need me to keep the lights on.