Machine Learning Engineer

Career.io is growing its machine learning team with a 100% remote, work-from-home Machine Learning Engineer role where you own ML products end to end. You will take work from problem definition through production and the metrics that prove impact. The focus spans data and entity resolution, retrieval, ranking, and matching, and applied LLMs and agentic systems, built with an AI-native development workflow. Engineers operate with autonomy in their domain and accountability for outcomes.

What you’ll own

• End-to-end product ownership from problem framing to production delivery to measurement of what changed.
• Experiments that earn their way into production, including taking responsibility for experiments that do not pan out and the decision to stop them.
• Canonical datasets for titles, companies, skills, and industries, including content-addressed IDs, faceted taxonomies, and alias graphs built up across tens of millions of rows.
• Rules-based resolution pipelines enhanced with LLM escalation, where the durable asset is the alias graph and escalation volume should decline over time.
• Nightly agent loops that adjudicate ambiguous entities, propose structural changes, and only commit after passing invariant checks and blast-radius limits.
• Job ingestion at scale across multi-source feeds, with deduplication, freshness, and attention to indexing economics.
• Job matching v2 using two-tower retrieval plus cross-encoder reranking, trained on outcome labels rather than clicks, with hard-negative mining, propensity weighting, and impression-time logging.
• Mobility embeddings learned from observed career sequences to capture similarity that a text encoder alone cannot recover.
• Pivot feasibility work to determine realistic moves, what is missing, and which intermediate roles worked for peers.
• Fine-tuning selectively when it earns its cost against outcome labels rather than for tasks handled well by a well-prompted frontier model.
• Agentic systems in production with human approval gates, where agents produce reviewable artifacts and execute only after human sign-off.
• Continuous skills inference from work artifacts rather than relying on static documents.
• New product surfaces where a correct outcome requires an LLM, and detection of areas where it does not.
• Evaluation infrastructure suitable for design review, including time-forward splits, calibration, offline-to-online agreement, and careful handling of feedback-loop degeneration and survivorship bias.
• Design under real constraints including GDPR, EU AI Act high-risk classification for employment AI, and client data commitments.

What you bring

• 5+ years shipping ML systems into production, with the ability to name the system, define the metric before and after, and explain how you knew the model caused the change.
• Strong depth in classical ML and deep learning applied to live products, not just notebooks and Kaggle datasets, using PyTorch or TensorFlow.
• Working fluency with LLMs in production: retrieval, evaluation, prompt and context engineering, plus judgment about when an LLM is the wrong tool.
• Experience shipping with agentic coding tools, specifically Claude Code, Claude Design, or close equivalents, and the ability to point to work you built with them.
• Software engineering fundamentals to own deployments end to end: Python, Git, AWS (we run AWS), containers, and the patience for messy, human-authored, self-reported data.

Technologies you’ll use

• PyTorch, TensorFlow
• Claude Code, Claude Design
• Python, Git
• AWS, containers

Preferred experience

• Entity resolution, record linkage, or taxonomy design at scale
• Ranking, recommendation, or two-tower retrieval systems
• Sequence models on longitudinal or event-stream data
• Embedding and vector retrieval systems in production
• Experiment design, causal inference, or off-policy evaluation
• Warehouse-native ML (dbt, Snowflake, or similar)
• Labor market, HR tech, or people-data domain experience
• Open-source contributions or publications