Build the future of payments with AI, Pix & Open Finance
Type: Full-Time
Reports to: CEO
The Company
Nova is building a new generation payment network (i.e. reinventing Visa/Mastercard) using AI on top of Pix and Open Finance.
Its first product is a B2C digital wallet that functions as an infrastructure bridge between traditional credit cards and Pix, allowing consumers to use their existing credit card limits anywhere, even to pay Pix and Boletos in installments.
The company was founded by an elite team of fintech veterans and has had top-tier venture capital investors (Monashees, Canary, Canaan and Propel) since its inception.
We recently raised a large Series A and are now assembling a small team of exceptional experienced professionals to build, launch and grow its next generation of products.
The Role
We are looking for an exceptional Purple Team Security Engineer to attack and defend Nova's payment network, working closely with the founders and the engineers who own each service. You will emulate the adversaries that target Brazilian fintechs, turn every finding into a detection or a control, and raise the security bar across the whole engineering team. We're looking for an outstanding security engineer who will.
Detect and respond
- Develop and tune detection rules across our cloud, Kubernetes, application and payment telemetry (Datadog logs, traces, audit trails, CloudTrail, EKS audit logs), based on relevant and emerging threats against fintechs in Brazil;
- Build and maintain detection-as-code pipelines: rules versioned in Git, tested against emulated attacks, deployed through CI, with coverage tracked against MITRE ATT&CK and D3FEND;
- Correlate threat intelligence (fraud rings, Pix scams, credential-stuffing campaigns, mobile malware) with internal telemetry to enrich detection logic and prioritize hardening;
- Lead security incident response end to end: triage, containment, forensics, and the post mortem that turns the incident into new detections and controls;
Attack and Harden
- Design and run adversary emulation against our own systems, using tools like Atomic Red Team, Caldera or custom tooling, focused on the attacks that matter in payments: account takeover, synthetic identity at onboarding, authorization/capture race conditions, replay and idempotency abuse, payout-before-capture failure paths;
- Harden the platform: IAM least privilege, secrets management, network segmentation, supply-chain and CI/CD controls, container and image hygiene;
- Review architecture and code for new services before they ship, with a threat model per service rather than a generic checklist, and coordinate external pentests and a responsible-disclosure program;
- Partner with the fraud and risk team so that fraud signals, device intelligence and onboarding controls are engineered as security controls, not only business rules;
Secure our AI, and use AI to secure us
- Red-team our AI agents and LLM-backed features: prompt injection (direct and indirect, through support conversations, documents, emails and web content), jailbreaks, tool-use and privilege abuse, data exfiltration through model outputs, and manipulation of agents that can move money or change account state;
- Design and enforce guardrails for authenticated AI actions: least-privilege tool scopes, input/output filtering, human-in-the-loop thresholds for high-risk actions, and audit trails that let us reconstruct what an agent did and why;
- Build detections for AI-specific abuse (injection attempts, anomalous tool-call patterns, agent sessions that deviate from policy) and run continuous adversarial evaluation of agents before and after each release;
- Use AI as a force multiplier in security operations: log triage, rule authoring, PoC generation, code review at scale, threat-intel summarization; build and maintain the security team's own agents and tooling;
- Set the secure-by-default patterns for how Nova engineers ship LLM features, aligned with OWASP Top 10 for LLM Applications and MITRE ATLAS;
Govern and Teach
- Map our controls to PCI DSS, LGPD and Banco Central requirements as Nova moves toward its own licenses, and own the technical evidence for audits;
- Prepare clear situation reports and quarterly security posture summaries for leadership: what was tested, what was found, what was fixed, what is still open;
- Teach and multiply: pair with backend engineers, run tabletop exercises, write walkthroughs and PoC demonstrations, and make the secure path the convenient path in our Go services, infrastructure and AI agents;
This is your chance to build a security program from the ground up at a payment network that handles other people's money and card data, as part of a select team at an ambitious and unique company.
The Qualifications
Required
- 5+ years in security engineering with real time on both sides: offensive engagements (internal red team, pentest, adversary emulation) and building or operating detection and response;
- Hands-on experience securing or attacking LLM-based systems: you have found and fixed prompt injection, tool-use abuse or data leakage in a real application, or built guardrails and evaluations for an agent that takes actions; you know OWASP Top 10 for LLM Applications and MITRE ATLAS and have opinions about where they fall short;
- Heavy daily use of AI in your own security workflow (log analysis, rule authoring, PoC development, code review, report writing) and the judgment to know when not to trust it;
- Proficiency in Python and Bash, and the ability to read and ship production code in a compiled language (our services are in Go; Go experience is a plus, not a gate);
- Hands-on experience ingesting, parsing and analyzing logs from diverse sources (cloud, Kubernetes, application, network, LLM traces) and writing detections in at least one query language (Datadog, Splunk SPL, KQL, Elastic DSL or similar);
- Familiarity with threat emulation and adversary simulation tools (Atomic Red Team, Caldera, PurpleSharp, AttackIQ or similar) and with MITRE ATT&CK as a coverage framework;
- Hands-on exploitation skills against web APIs and backend services: authentication and session flaws, authorization bypass, injection, race conditions, business-logic abuse in money-moving flows;
- Cloud and container security in practice: AWS IAM, VPC and network controls, Kubernetes RBAC and pod security, secrets management, CI/CD hardening;
- Experience leading incident response on a system that handles customer money or data, including the post-incident write-up, applying frameworks such as NIST 800-61;
- Experience working in fast-paced environments, balancing high quality with speed;
- Full professional proficiency in English;
Preferred
- Experience in fintech or payments in Brazil: card authorization/capture, acquiring, Pix, Open Finance, chargebacks and dispute flows;
- Experience taking a company through PCI DSS, SOC 2 or a Banco Central authorization process, or building the controls for one; working understanding of LGPD;
- Fraud and identity: device fingerprinting, bot and automation detection, KYC/onboarding abuse, synthetic identity;
- Experience building AI agents or LLM tooling yourself (agent frameworks, MCP or function-calling integrations, evaluation harnesses); public AI security research, CTF results or contributions to tools like garak, PyRIT or promptfoo;
- Experience with Datadog Cloud SIEM, Workload Security, LLM Observability or equivalent cloud-native tooling;
- Certifications such as OSCP, OSEP, GCIH, GCFA or CRTO; a public track record (CVEs, bug bounty, talks, open-source tooling) counts just as much
- Background in early-stage startups, especially as the first security hire.
The Target Companies
Candidates with at least 2 years of full-time experience in the companies below will be prioritized:
- Strong Security Companies and Consultancies: Tempest, Conviso, Clavis, Trail of Bits, NCC Group, Mandiant, CrowdStrike, Datadog Security Research;
- Big Tech: Google, Amazon, Apple, Meta, Microsoft
- Other Tech: Stripe, Cloudflare, Uber, Airbnb, Netflix, Spotify, Snap, X (Twitter)
The Selection Process
Materials about Nova, including a pitch deck and pitch videos, will be made available through the process.
- Screening Call: 15 min call with HR for alignment on requirements and qualifications;
- Deep Dive Interview: 60 min call with the Hiring Manager to deep dive on past experiences and profile, including one offensive engagement you ran and one detection or control you built from it;
- Technical Exercise: take-home or live, your choice. Threat-model a simplified card-funded Pix payout flow, propose the top three detections and the top three hardening changes, and show how you would validate one detection with an emulated attack;
- Bar Raiser Interview: 45 min call with another leader to validate assessment with a 2nd opinion
- Cultural Fit Interview: 30 min call with the CEO to evaluate culture fit and answer questions about Nova
The Offer Package
- Model: PJ;
- Other Benefits: premium health insurance, flexible hours, paid vacations and Wellhub.