Head of AI

Head of AI

PITZ | Reports to CEO

The short version

PITZ is building the operating system for the $1T global automotive aftermarket. We're not another AI wrapper — we're an AI-native infrastructure company where voice is the interface, real-time decisions drive revenue, and every mechanic interaction makes the system smarter.

We need someone to own the intelligence layer. Not to research. To build, ship, and scale AI systems that run a real supply chain across three countries.

What PITZ actually is

The automotive aftermarket is where cars get fixed after the warranty ends. It's a trillion-dollar industry that still runs on phone calls, guesswork, and handwritten notes.

PITZ changes that. A mechanic describes a problem — in Mexican slang, Brazilian Portuguese, or English — and our system understands the intent, predicts the failure, identifies the exact part from 1.4M+ SKUs, checks availability across integrated sellers, and triggers an order. In seconds.

We're live in Mexico and Brazil with 1,400+ workshops. Entering the US in Q4 2026.

Our architecture has three layers:

Decision Layer (AI-native) — Voice → Intent → Action. Multilingual voice inference trained on mechanic slang, ambient noise, and regional failure patterns. Real-time decision engine that determines what to fix, what to order, and from whom. Continuous learning from live operations.

Coordination Layer (System of Record) — One shared operational graph connecting orders, inventory, pricing, payments, workshops, sellers, and credit. SLA-aware execution that knows who fulfills, when, and at what margin.

Execution Layer (Market Embedded) — Revenue orchestration for shops. Supply and fulfillment for sellers. Embedded credit and incentives that increase ticket size 35-50%.

You own the Decision Layer. You shape the Coordination Layer. You make sure the Execution Layer gets smarter with every transaction.

What you'll do

Build and scale the voice-to-action pipeline. Our voice system processes mechanic speech in noisy shop environments, across multiple languages and regional dialects. You'll own the full pipeline: speech recognition → intent extraction → diagnostic prediction → SKU mapping → order optimization.

Architect the real-time decision engine. Every repair creates a decision: what part, from which seller, at what price, with what delivery SLA. You'll build the intelligence that makes these decisions faster and more accurate than any human could — then prove it with data.

Design the credit scoring models. PITZ Capital uses behavioral and operational data to underwrite mechanics who have no traditional credit history. You'll build the scoring engine that turns 18M+ diagnostic interactions into underwriting signals.

Scale AI across three markets simultaneously. Mexico is mature. Brazil is scaling. The US launches Q4 2026 with integrations into Identifix, ALLDATA, OBD platforms, Mitchell1, and PartsTech. You'll adapt models to each market's vehicle fleet, parts ecosystem, and language patterns.

Build the data flywheel. Every voice interaction, every order, every delivery, every payment creates training data. You'll design the systems that capture, clean, and feed this data back into models that compound in accuracy over time.

Ship, don't research. We need production AI systems that handle thousands of daily interactions reliably. Research is a tool, not the job. If a simpler model ships faster and works, we ship the simpler model.

What you bring

8+ years building AI/ML systems in production. Not prototypes. Not notebooks. Systems that handle real users, real noise, real edge cases at scale.

Deep experience with voice/NLP in messy, real-world environments. Bonus if you've worked with multilingual systems, domain-specific language, or noisy audio inputs. Academic NLP experience alone won't cut it — we need someone who's dealt with speech that doesn't look like a textbook.

Strong engineering instincts. You can architect a system, write the code, review a pull request, and debug a production issue. You don't need a team of 20 to ship something meaningful.

Experience with recommendation systems, decision engines, or marketplace optimization. The core of our product is matching — the right part, from the right seller, at the right price, at the right time. If you've built systems that make these kinds of decisions at scale, you'll move fast here.

Comfort operating across the full stack. Model training, data pipelines, inference optimization, API design, monitoring, cost management. You'll have a small team, and the scope is wide.

Startup DNA. You've worked in environments where speed matters more than perfection. You know when to build vs. buy. You can operate with ambiguity, make decisions with incomplete data, and course-correct fast.

Nice to have

* Experience with embedded credit/fintech scoring models

* Spanish or Portuguese language skills (not required, but valuable)

* Background in supply chain, logistics, or marketplace dynamics

* Experience scaling AI systems across multiple geographies

* Familiarity with automotive or industrial vertical data

What you won't find here

* A research lab. We ship to production, not to arXiv.

* A feature factory. You'll own the entire intelligence layer and its roadmap.

* A "Head of AI" title with no real authority. You report directly to the CEO and your decisions shape the product.

* Politics. We're a 30-person company building infrastructure for a trillion-dollar market. There's no time for anything that doesn't move the business forward.

Why this matters

Most AI roles are about making an existing product slightly better. This one is different.

At PITZ, AI isn't a feature — it's the business. Voice is our primary interface. Decisions run through our models. Credit depends on our scoring. The entire company compounds on the intelligence layer you build.

You'll work with proprietary data that no other company has: 18M+ real diagnostic interactions from mechanics across Latin America, mapped to SKUs, orders, payments, and outcomes. This data moat grows with every transaction. Your models get better every day — not because of more compute, but because of more reality.

And the timing is right. We're entering the US market, raising our Series A, and building toward an IPO within four years. The decisions you make in the next 12 months will define the company for the next decade.