ML/AI Engineer | EBANX Global Negotiation Guide

Negotiation DNA: Private ($3.5B valuation) Agentic Commerce LatAm Payments Pix Integration Beyond Borders 2026 Know Your Agent Agent Modeling Payment Intelligence

---

Region	Base Salary	Stock (Options/4yr)	Bonus	Total Comp
Curitiba	R$228,000-R$348,000 / $45,000-$68,000	R$90,000-R$200,000 / $18,000-$39,000	R$25,000-R$50,000 / $5,000-$10,000	R$343,000-R$598,000 / $68,000-$117,000
São Paulo	R$276,000-R$408,000 / $54,000-$80,000	R$115,000-R$250,000 / $23,000-$49,000	R$32,000-R$62,000 / $6,000-$12,000	R$423,000-R$720,000 / $83,000-$141,000
Austin TX	$160,000-$220,000	$45,000-$100,000	$18,000-$32,000	$223,000-$352,000

---

Negotiation DNA

ML/AI Engineers at EBANX build the intelligence layer of a payments platform that processes $1B+ in monthly TPV for merchants like Spotify, Uber, AliExpress, and Shein across 15+ countries in Latin America and Africa. Their models power fraud detection, transaction routing optimization, conversion rate prediction, currency risk management, and merchant analytics. The Feb 10, 2026 Beyond Borders study fundamentally elevates the ML/AI Engineer's role: agentic commerce means that AI agents — not just human consumers — will initiate cross-border payments, and EBANX needs ML models that can understand, predict, and validate agent behavior in real time. This is the most direct connection between a role and EBANX's strategic future. ML/AI Engineers who can build agent behavior models for financial transactions are among the most sought-after hires in global payments.

As a private company valued at $3.5B with Advent International and FTV Capital backing, EBANX offers options with 4-year vesting that carry significant upside for ML/AI Engineers who are building the core intelligence of the platform. The Know Your Agent (KYA) initiative is, at its core, an ML problem: building models that verify agent authorization, detect anomalous agent behavior, and score the risk of agent-mediated transactions in real time — all while processing Pix payments with sub-second latency. Agentic Testing also has a deep ML dimension: using ML models to generate synthetic agent behavior for testing, predicting which agent-payment-method-country combinations are most likely to fail, and building reinforcement learning systems that continuously improve test coverage. Curitiba houses the core ML engineering team; São Paulo focuses on merchant-facing ML products; Austin supports US-market ML applications and global model governance.

---

Level Mapping:

EBANX	Google	Meta	Stripe	dLocal	Nubank
ML/AI Engineer	L4 MLE	IC4 MLE	ML Engineer	ML Engineer	ML Engineer
Senior ML/AI Engineer	L5 MLE	IC5 MLE	Senior MLE	Senior MLE	Senior ML Engineer
Staff ML/AI Engineer	L6 MLE	IC6 MLE	Staff MLE	Staff MLE	Staff ML Engineer

---

Agentic Commerce — The Know Your Agent Premium

1. Beyond Borders Agent Intelligence

The Feb 10, 2026 Beyond Borders study positions ML/AI as the enabling technology for agentic commerce. You are the person who builds the intelligence layer:

"The Beyond Borders study from Feb 10 makes it clear that agentic commerce is EBANX's next growth frontier. As an ML/AI Engineer, I would build the models that make this possible — agent behavior prediction, autonomous transaction risk scoring, and real-time agent authorization verification. This is the core intelligence layer of EBANX's agentic commerce platform. I've built production ML systems for financial transaction processing. For building EBANX's agent intelligence, I'm targeting R$330,000 base in Curitiba with R$185,000 in options over four years."

2. KYA Behavioral Modeling

Know Your Agent depends on ML models that can distinguish legitimate agent behavior from fraudulent or unauthorized activity:

"Know Your Agent at machine speed means building ML models that can verify agent authorization, detect behavioral anomalies, and score transaction risk — all within the latency budget of a Pix payment. The model needs to work across 15+ countries with different payment methods, regulatory frameworks, and consumer behavior patterns. I've built real-time behavioral anomaly detection models for financial systems. For building EBANX's KYA behavioral models, I'm asking for total comp of R$650,000+ in São Paulo, with an options grant that reflects the pre-liquidity upside and the strategic nature of KYA."

3. Pix Agent Risk Scoring

Pix's instant settlement means ML risk scoring for agent transactions must run in single-digit milliseconds:

"Pix settles in real time, which means an ML risk score for an agent-initiated Pix transaction must be computed in single-digit milliseconds. This is an inference optimization challenge combined with a model accuracy challenge — you need to be fast and right. I've optimized ML inference pipelines for real-time financial decisions. My ask is $210,000 base in Austin with $90,000 in options over four years."

4. Agentic Testing with ML

ML can generate synthetic agent behavior for testing and predict which agent-payment combinations are riskiest:

"Agentic Testing can be dramatically improved with ML: generative models that create synthetic agent behavior patterns for testing, predictive models that identify which agent-payment-method-country combinations are most likely to fail, and reinforcement learning systems that optimize test coverage. I want to build ML-powered Agentic Testing for EBANX. For this scope, I'm targeting R$340,000 base with R$190,000 in options in Curitiba."

---

Negotiate Up Strategy: Target a 15-20% uplift on the initial base offer. For Curitiba, push for R$325,000+ base with R$175,000+ in options (4yr vest). For São Paulo, target R$380,000+ with R$220,000+ in options. For Austin, target $205,000+ with $85,000+ in options. Your accept-at floor: R$255,000 / R$300,000 / $170,000 base respectively, with minimum options of R$100,000 / R$130,000 / $50,000. ML/AI Engineers should also negotiate for GPU compute budget (critical for model training at a private company), ML conference attendance (NeurIPS, ICML), and publication rights for non-proprietary research — these non-cash perks can be worth $10,000-$20,000 annually and signal the company's commitment to ML engineering as a first-class discipline.

---

Evidence & Sources:

1. EBANX Beyond Borders 2026 Study (Feb 10, 2026) — Agentic commerce ML requirements and agent behavior modeling for LatAm payments
2. Glassdoor and Levels.fyi — ML/AI Engineer compensation at Brazilian fintechs (Nubank, Stone, iFood) and US payments/ML companies (Stripe, PayPal, Google)
3. EBANX corporate data — $3.5B valuation, Advent International and FTV Capital backing, $1B+ monthly TPV, ML-powered fraud detection and routing optimization
4. Banco Central do Brasil Pix performance requirements (2025-2026) — Real-time inference latency constraints for Pix-integrated ML systems
5. AI/ML Compensation Report by Rora Negotiation (2026) — ML engineer salary benchmarks for fintech companies in LatAm and US markets