AI/ML Engineer — MCP Infrastructure

Project Brief We're building out a Model Context Protocol (MCP) infrastructure for a mid-market US software company that's moving fast into agentic AI. This is a new, independent engineering team — separate from existing projects — building from the ground up. Your role has a strong DevOps lean: keeping the MCP infrastructure healthy, observable, and scalable as agents and tools get added. You'll own deployments, pipelines, and production stability. The title is flexible — what matters is the function: someone who can hold the technical ground on infrastructure while the team builds. Must-Haves Strong DevOps fundamentals in practice — you own deployments, not just contribute to them. Container orchestration (Docker + ECS/Fargate or equivalent), GitHub Actions for CI/CD, AWS (ECS, RDS, S3, CloudWatch), and MCP protocol familiarity are part of your daily work Hands-on production experience with LangChain, LangSmith, and/or LangGraph — not course projects or prototypes LLM infrastructure thinking — you understand token cost, latency tradeoffs, rate limits, and how to monitor them in production Comfortable working autonomously in a small team without daily hand-holding Near-native English — daily async communication with a US-based technical lead and client stakeholders Nice to Have LangSmith tracing and evaluation features — setting up traces, running evals, interpreting results Experience collaborating directly with a client-side senior engineer — comfortable integrating into an established technical dynamic and contributing without needing to redefine it Familiarity with observability tooling beyond CloudWatch — Datadog, Grafana, or similar What You Will Do Monitor, maintain, and optimize the agentic infrastructure running on LangChain / LangSmith / LangGraph Manage container-based deployments and ensure stability across environments Build and own CI/CD pipelines for agent and model deployments Set up and maintain observability — tracing, alerting, and performance dashboards for LLM-based systems Support MCP server integration as the client-side team ships new components Identify and resolve latency, cost, and reliability issues before they become production incidents Work closely with the client's MCP technical lead — small team, no bureaucracy, your infrastructure decisions are immediately visible Why This Could Be Your Next Big Move Infrastructure that actually matters — You're not maintaining a toy. This is a production agentic system with real users, real costs, and real consequences when things break. Observability as a craft — LLM systems fail in non-obvious ways. You'll build the tooling that makes invisible problems visible before the client notices them. Direct access to the technical decision-makers — Small team, no bureaucracy. Your work is immediately visible to the client's lead engineer and a senior technical advisor MCP is the frontier — Model Context Protocol is where enterprise AI is heading. You'll have production experience on it before most engineers have even read the spec. Benefits & Compensation $4000 - $5000/month — paid in USD, bi-weekly via Deel US Eastern Time hours (EST) — Monday to Friday, 9:00 AM–6:00 PM EST Fully Remote — work from anywhere in Latin America Long-term contract — starting with a 6-month contract, with potential to extend Paid PTO — accrual begins after 3-month trial period Referral Program — earn a bonus for referring talent that gets hired About OneSeven Tech OneSeven Tech is an AI-enabled software agency headquartered in Miami, with 9 years building AI-driven digital products for scaling and established companies. We specialize in consulting, designing, and developing ROI-focused solutions across 22 industries — from healthcare and legal to real estate and fintech. Our multidisciplinary team blends AI strategy, UX/UI, web and mobile engineering, and machine learning to deliver real outcomes. With 101+ successful projects completed and a 4.9/5 client rating, we work as a deeply embedded partner — not just a vendor. We operate as a remote-first organization with a global team and a US-based client portfolio. #J-18808-Ljbffr OneSeven Tech