Full Stack Engineer

Role Overview

The Full Stack Engineer is responsible for developing and managing features across Qashio's corporate card and spend management platform, encompassing user interfaces, APIs, backend services, and data structures. Reporting to the CTO, this position holds significant influence over both the development process and end products. This role is deeply integrated with AI technology, requiring proficiency in AI-assisted coding tools to safely and efficiently deliver AI-enhanced features within a regulated financial context.

Responsibilities

• Design, build, and maintain complete full-stack features using React, Next.js, Node.js, NestJS, APIs, and data models, ensuring accountability for production stability.
• Create and manage RESTful APIs and integrations with internal systems and third-party services, including card issuers, banking, ERP, and accounting platforms.
• Translate business and product requirements into technical designs and communicate trade-offs effectively with stakeholders from product, finance, and operations.
• Write well-structured, documented, and tested code, keeping changes small for effective review and ownership.
• Troubleshoot and resolve production issues while participating in on-call rotations and incident responses for owned services.
• Ensure all deliverables comply with security, privacy, and regulatory standards, especially concerning authentication, authorization, payment processes, and sensitive financial and personal data.
• Continuously improve development workflows, tooling, and CI/CD pipelines.
• Leverage agentic AI coding tools such as Claude Code, Cursor, and GitHub Copilot to streamline delivery by breaking down tasks, drafting specifications and tests, and guiding AI-generated implementations.
• Review and rigorously verify AI-generated code, maintaining ownership of the quality and security of every line before deployment.
• Maintain the AI tooling context by updating repository instructions, reusable prompts, internal connections, and current architecture documentation.
• Elevate team proficiency with AI tools by sharing effective workflows, discontinuing ineffective ones, and establishing internal standards for safe AI-assisted development.
• Develop AI-driven product features using APIs from OpenAI, Anthropic, AWS Bedrock, implementing safeguards, performance monitoring, and controlling costs and latency.

Qualifications and Experience

• Bachelor's degree in Computer Science, Engineering, or a related discipline, or equivalent practical experience.
• Over 5 years of experience building and maintaining production web applications as a full-stack engineer with substantial expertise in both frontend and backend development.
• Frontend skills include React, Next.js, TypeScript, and JavaScript.
• Backend expertise with Node.js, NestJS, TypeORM, and TypeScript.
• Proficient with relational databases like PostgreSQL and at least one NoSQL database such as MongoDB, DynamoDB, or Cassandra.
• Experience with AWS cloud services, containerized environments, automated testing, CI/CD pipelines, and production monitoring.
• Skilled in RESTful API design, microservices architecture, and API gateway integration.
• Strong understanding of modern software design patterns and secure coding practices applicable to financial and sensitive data systems.
• Hands-on use of agentic coding tools (Claude Code, Cursor, GitHub Copilot, etc.) in professional production environments, with ability to articulate successes and challenges.
• Expertise in context engineering for AI tools, enabling meaningful AI output and discerning when manual coding is necessary.
• Disciplined in test-first development, incremental reviews, version checkpointing, and quick rollback procedures.
• Deep awareness of security risks related to AI-generated code and effective mitigation techniques including static and dependency analysis, secret scanning, and human review controls.
• Clear understanding of data privacy, identifying what customer or cardholder information can be safely processed by AI models under regulation.
• Preferred: Experience with LLM API integration (OpenAI, Anthropic, AWS Bedrock), including handling streaming, retries, token management, caching, and graceful degradation.
• Familiarity with retrieval-augmented generation, vector search, structured outputs, and tool/function calls.
• Experience assessing LLM system quality, building evaluation datasets, identifying quality declines, and setting acceptance criteria for probabilistic outputs.
• Knowledge of OWASP Top 10 security issues for LLM applications and their mitigations.
• Prior working knowledge of fintech or payments sector compliance standards such as PCI DSS, SOC 2, and data residency.

Essential Competencies

• Judicious decision-making in accepting, correcting, or discarding AI-generated output, recognizing when human-initiated problem solving is required.
• Able to maintain high review standards quickly, serving as both author and reviewer effectively.
• Security-first mindset, rigorously verifying code especially where sensitive money, credentials, or customer data are involved.
• Fast adoption and replacement of tooling as workflows evolve.
• Strong ownership and autonomy, working confidently with ambiguity, proactively making decisions, and escalating appropriately when needed.
• Clear and precise written communication for specifications, tickets, and code reviews.
• Solid understanding of user needs, and competence in translating business goals into technical implementations.
• Analytical problem-solving skills and data-driven decision-making abilities.
• Collaborative approach to working with cross-functional teams including product, design, operations, and compliance.

Additional Information

At Qashio, AI-assisted engineering is integrated into the standard development process rather than being experimental. Engineers are expected to routinely use AI tools, critically assess their output, and are accountable for the final quality of shipped products. Performance evaluation focuses on delivered results and avoiding defects rather than manual coding volume.