SLM Trainer

JOB SUMMARY

Our client is bringing an existing product in-house that has, to date, been built and managed by a US-based development partner. As part of a small in-house pod based in the Philippines, we’re looking for a mid-level SLM Trainer to own the day-to-day training, fine-tuning, and evaluation of the Small Language Models (SLMs) that power this agentic AI product, working under the direction of the Senior Agentic AI Lead.

This is a hands-on execution role, not a research or prototype position. The product is a governance system used by global enterprises, so you’ll be working on a live, production model — with real users, real data, and real consequences for regressions. You’ll be responsible for keeping the SLMs accurate, safe, and performing to spec, while documenting your work clearly enough for the wider pod and the client to follow.

JOB RESPONSIBILITIES

• Train, fine-tune, and continuously improve the product’s Small Language Models (SLMs) under the direction of the Senior Agentic AI Lead
• Build and maintain evaluation and regression-test suites to catch model drift or degraded performance before it reaches production
• Monitor live model behavior in production, triage anomalies, and escalate incidents appropriately
• Curate, clean, and manage training and evaluation datasets in line with governance and data-handling standards
• Work within the Claude Code, Postgres, and Rust stack to support model integration and testing
• Document model versions, training decisions, evaluation results, and known limitations
• Collaborate closely with the junior developer on the pod to keep documentation and test coverage current
• Manage your own workstream through Jira and contribute to the shared GitHub codebase
• Flag risks around model behavior, bias, or data handling to the Senior Agentic AI Lead promptly
  
  

QUALIFICATIONS

• Hands-on experience training and fine-tuning Small Language Models (SLMs) for a live, production system — not just experimentation or prototype work
• Solid understanding of model evaluation, monitoring, and regression-testing practices
• Familiarity with agentic AI systems and how SLMs are used within them
• Working knowledge of data handling and governance practices — access controls, data quality, versioning
• Comfortable with Python and standard ML/data tooling
• Some exposure to Claude Code, Postgres, or Rust (or a clear ability to ramp up quickly)
• Security-conscious approach to model training and data handling
• Strong documentation habits and clear written communication — able to explain model behavior to non-technical stakeholders
• Familiarity with GitHub and Jira
• Mid-level experience — enough independent judgment to own a workstream day-to-day, without requiring full architectural ownership
• Comfortable taking direction from a senior lead while flagging issues and risks proactively