Machine Learning Engineer

## About Lens Lens is building credibility infrastructure for the information economy — an AI-powered Media Objectivity Score (IQS) that makes influence visible, measurable, and auditable. Backed by $80K pre-seed and seeking a $3M seed round, we are a lean, mission-driven team building the Bloomberg Terminal for information. ## The Role We are looking for a Machine Learning Engineer to own the intelligence behind the IQS. The IQS is the core of everything Lens builds — a real-time, multi-dimensional credibility scoring system that is transparent, auditable, and reproducible. You will build and iterate on the ML systems that make this possible. ## What You'll Do - Architect and implement the IQS scoring pipeline: prompt design, LLM API orchestration, and result aggregation - Build multi-agent verification workflows to improve scoring accuracy and catch edge cases - Design and implement score reproducibility systems: content hashing, cache keying by model version and prompt - Develop and maintain fine-tuning pipelines using RLAIF on expert-validated article datasets - Collaborate with the Expert Panel to translate human credibility judgments into training signals - Build abuse-handling and adversarial robustness features into the scoring architecture - Evaluate model performance across IQS dimensions and produce calibration reports - Progress the system from frontier LLM APIs (Phase 1) toward in-house fine-tuned models (Phase 3) ## What We're Looking For - 3+ years of applied ML or NLP engineering experience - Strong familiarity with large language models and prompt engineering - Experience building production inference pipelines (Anthropic, OpenAI, or equivalent APIs) - Understanding of evaluation frameworks, benchmarking, and human-in-the-loop validation - Proficiency in Python and ML tooling (Hugging Face, PyTorch, or equivalent) - Ability to communicate technical design decisions to non-technical stakeholders ## Nice to Have - Experience with RLHF or RLAIF fine-tuning workflows - Familiarity with multi-agent architectures (LangGraph, CrewAI, AutoGen, or equivalent) - Background in computational linguistics, journalism studies, or information science - Experience with extended-thinking or chain-of-thought model APIs ## Tech Stack - Python - Anthropic / OpenAI APIs - Hugging Face Transformers - Multi-agent frameworks