Data Scientist

This is a fully remote position, open to applicants in Philippines.

📋 Description

• Design, construct, and assess LLM agents that interact with actual dealership systems — scheduling service appointments, addressing vehicle availability inquiries, resolving customer identity issues, and escalating matters to humans with complete context.

• Take ownership of the tool-utilization layer: specify the tools and function schemas that agents invoke, establish guardrails for each, and define system behavior when a downstream service is slow or unavailable.

• Create agent evaluation infrastructure — including offline evaluation sets, adversarial and edge-case suites, live A/B testing, and regression gates to prevent deployment during quality declines.

• Develop and implement escalation logic and confidence thresholds: determining when the agent acts, when it confirms, and when it hands off to a human.

• Build and maintain RAG systems over dealership content (service history, OEM documentation, policies, inventory) using Bedrock embeddings, pgvector on Aurora, and OpenSearch Serverless.

• Manage prompt architecture, versioning, and change control as a vital engineering artifact kept under source control.

• Construct, validate, and deploy predictive models on lakehouse data — including gross profit forecasting, customer lifetime value, defection risk, next-service prediction, identity resolution, and inventory pricing signals.

• Oversee the complete model lifecycle: feature engineering, training, validation, deployment, monitoring, and retraining.

• Release models with drift and degradation monitoring from day one.

• Translate operational and ownership business questions into well-defined modeling problems and challenge when a query would be a more appropriate solution than a model.

• Measure and convey model impact in dealership terms: gross profit, units sold, retention, CSI, and labor hours saved.

⛳️ Requirements

• Over 4 years of experience applying data science in production, with models that have influenced real decisions and outcomes.

• Proficient in Python and SQL. You write code that others can execute and maintain.

• Practical experience in building LLM agents with tool usage and function calling — not limited to prompt engineering.

• Be prepared to discuss a system you developed and how you assessed its performance.

• Hands-on RAG experience: embeddings, vector search, chunking, retrieval evaluation, and re-ranking.

• Solid understanding of statistical principles and a genuine approach to uncertainty. We favor a well-calibrated interval over a confident point estimate.

• Experience deploying models into production — not merely passing notebooks to an engineering team.

• Comfortable working with AWS ML tools (Bedrock, SageMaker, Lambda) and within a lakehouse or data warehouse environment.

🏝️ Benefits

• Fully remote – work from anywhere in the Philippines.

• Engage with live agentic AI systems — not proof of concepts, not presentations, and no handoffs to engineering.

• Data infrastructure is already established (AWS lakehouse, Databricks), allowing you to concentrate on modeling and agents rather than plumbing.

• Join a small, senior, high-autonomy team that fosters a documentation-first culture.

• Opportunity to define the evaluation and deployment standards that every new model and agent will adhere to.

• Be part of a team culture that values intellectual honesty, technical expertise, and follow-through.