Sr AI Engineer- Voice & Agentic AI

The Senior AI Engineer will design, build, and ship production-grade AI systems, with a focus on real-time, agentic, and voice applications where reliability and latency are non-negotiable. This person owns AI features end to end: from agent orchestration and retrieval through deployment, evaluation, and monitoring in live environments. They work autonomously with minimal supervision while collaborating closely inside an embedded engineering team, and are expected to raise risks early, communicate proactively, and ship to a high bar.

The heart of this role is making intelligent systems work in the real world, where an agent answering a customer has to respond fast, stay accurate, and never give a confidently wrong answer. You will bridge AI engineering and production software: building agent workflows, fast and accurate retrieval, clean APIs, and the evaluation discipline that keeps it all trustworthy as it scales across customers.

What are the main responsibilities of this role?

• Design and build LLM-powered agents and multi-step workflows using orchestration frameworks (LangGraph or equivalent), with explicit, auditable state.
• Build retrieval (RAG) systems that are both accurate and fast enough to meet strict real-time latency budgets; making deliberate, defensible tradeoffs between recall and response time.
• Own AI features end to end: from prototype to production deployment, monitoring, and iteration.
• Refactor and improve live production systems without disrupting active customers.
• Build and maintain evaluation and regression harnesses so changes are validated against scenarios, not demos; and bugs are caught before they reach production.
• Create scalable Python APIs and services that expose AI functionality to web, voice, and mobile clients.
• Implement observability, tracing, and performance monitoring across deployed AI features - tracking latency, quality, and cost.
• Work across a multi-LLM stack (OpenAI, Anthropic, Google) behind a provider-abstraction layer, keeping model choice a configuration decision rather than a rewrite.
• Collaborate daily in a remote, sprint-based team, picking up scoped work, reviewing PRs through quality gates, and surfacing risks and blockers early.

Which skills and experience do we envision for this role?

• 4+ years in AI/ML or AI-focused software engineering, with real products shipped to production.
• Hands-on experience building LLM applications and agents (OpenAI, Claude, open-source models) using agent/orchestration frameworks (LangGraph, LangChain, or equivalent).
• Strong RAG and retrieval experience - embeddings, vector databases (Pinecone, Weaviate, pgvector, or similar), and the practical skill of tuning retrieval for both quality and speed.
• Solid Python, with experience building production APIs (FastAPI or similar).
• Experience evaluating and testing AI systems (RAGAS, Maxim, Promptfoo, LLM-as-judge, or equivalent) and a "done means tested, not demoed" mindset.
• Production experience on cloud (AWS preferred) with Docker, CI/CD, and infrastructure-as-code (Terraform).
• Strong software fundamentals: automated testing, code review, version control (Git), and modular design.
• Demonstrated experience deploying, monitoring, and refactoring AI systems running in production.
• AI-native development workflow (Cursor, Claude Code, Copilot) and the discipline to ship AI-assisted code safely through review and automated gates.
• Excellent written and spoken communication, strong autonomy, and comfort working in a remote team with at least 4 hours/day overlap with US Eastern time.

Nice to have

• Real-time or low-latency voice/conversational systems experience (streaming STT/TTS pipelines, turn-taking, barge-in).
• Familiarity with telephony or real-time media infrastructure (Twilio, LiveKit, WebRTC, Deepgram, ElevenLabs).
• Multi-tenant or configuration-driven platform experience.
• Knowledge-base ETL / data ingestion pipeline experience.
• Exposure to AI guardrails, governance, and compliance in regulated domains.

Join our team and enjoy:

• Access to Rootstrap University,Conferences/Certifications, and a Mentorship Program for your professional growth.
• Learning Bonus
• Opportunities to organize cross-functional initiatives and receive 360 Feedback to improve your skills continuously.
• The flexibility to work remotely or from our offices in Montevideo, Buenos Aires, and Medellín, with a flexible time schedule and Workation program.
• Gym benefits, psychological counseling, and weekly lunch reimbursements with special foods in the offices.
• An Onboarding kit and access to cutting-edge technologies and tools to make your work easier.
• After offices, Prizes, and special occasions gifts to celebrate your achievements.
• People Care referent to support your well-being and personal development.

How we work:

• We offer a flexible and diverse work environment that fosters multicultural talent
• We value autonomy and creativity, We want people to bring ideas to the table and to think as leaders.
• We put our focus on real human connections
• We have a commitment to excellence

We embrace an AI-first mindset, continuously exploring how artificial intelligence can enhance the way we build products, solve problems, and deliver value. Across all stages of the development process, we encourage our teams to actively leverage AI to boost productivity and drive innovation.

With offices in Uruguay, Argentina, and Colombia, we offer the flexibility to work remotely. Discover more about life at Rootstrap at https://www.rootstrap.com/careers.

All of our candidate searches are covered under Uruguayan Law 19691 (the purpose of this law is to promote the inclusion of people with disabilities).

As part of our Security and Compliance process, all seniority levels and roles must follow the company’s information security policies and secure development practices, and promptly report any suspected security incidents or vulnerabilities.