About the role
structured by ORIABOUT THE ROLE You’ll be the technical founder driving the machine learning and AI backbone behind Known — an intelligent, compatibility-driven dating platform that blends psychology, data, and human-like conversation. You’ll design and ship the systems that make Known feel magical: personalized matching algorithms,…
What you will do
- Design and implement multi-stage matching systems (embedding-based retrieval + LLM re-ranking) for compatibility scoring, search, and personalization.
- Develop and maintain ML pipelines for data ingestion, feature generation, model training, evaluation, and inference.
- Prototype and productionize agentic workflows for natural-language and voice interactions (e.g., AI-assisted intake interviews, voice matching, or conversation agents).
- Deploy and monitor ML models in production with guardrails for performance, fairness, and safety.
- Run offline & online experiments (A/B and multivariate) to measure real-world outcomes such as engagement, match success rate, and conversation quality.
What they are looking for
- 3+ years in applied ML or data science engineering roles, ideally working on recommendation, search, or personalization systems.
- Strong proficiency in Python and modern ML frameworks (PyTorch, TensorFlow, JAX, Hugging Face).
- Experience with LLMs, embeddings, and agentic workflows.
- Understanding of A/B testing and human-in-the-loop system design for model evaluation in production.
- You care about building safe, fair, and human-centered AI experiences.
Nice to have
- Familiarity with ANN search systems and modern MLOps tools is a plus.
- Reinforcement learning or preference modeling experience is a strong plus.
Full posting text
ABOUT THE ROLE
You’ll be the technical founder driving the machine learning and AI backbone behind Known — an intelligent, compatibility-driven dating platform that blends psychology, data, and human-like conversation. You’ll design and ship the systems that make Known feel magical: personalized matching algorithms, adaptive recommendation loops, and natural voice/LLM-based interactions that help users connect meaningfully.
You’ll work closely with the founding team (product, platform, and design) to shape both the data and ML foundations and the user-facing experiences that differentiate Known. This is a hands-on role with ownership across research, prototyping, and production deployment.
RESPONSIBILITIES
- Design and implement multi-stage matching systems (embedding-based retrieval + LLM re-ranking) for compatibility scoring, search, and personalization.
- Develop and maintain ML pipelines for data ingestion, feature generation, model training, evaluation, and inference.
- Prototype and productionize agentic workflows for natural-language and voice interactions (e.g., AI-assisted intake interviews, voice matching, or conversation agents).
- Deploy and monitor ML models in production with guardrails for performance, fairness, and safety.
- Run offline & online experiments (A/B and multivariate) to measure real-world outcomes such as engagement, match success rate, and conversation quality.
- Collaborate cross-functionally with platform engineers and product designers to integrate AI seamlessly into the Known user experience.
REQUIREMENTS
- 3+ years in applied ML or data science engineering roles, ideally working on recommendation, search, or personalization systems.
- Strong proficiency in Python and modern ML frameworks (PyTorch, TensorFlow, JAX, Hugging Face).
- Experience with LLMs, embeddings, and agentic workflows.
- Understanding of A/B testing and human-in-the-loop system design for model evaluation in production.
- Familiarity with ANN search systems and modern MLOps tools is a plus.
- Reinforcement learning or preference modeling experience is a strong plus.
- You care about building safe, fair, and human-centered AI experiences.
EXAMPLE PROJECTS
- Develop a user matching system based on profile information, onboarding transcripts and engagement behavior.
- Build a dynamic profile enrichment pipeline that integrates behavioral and linguistic features into user representations.
- Deploy a lightweight LLM-powered voice agent for user intake and conversational matchmaking.
- Create an evaluation harness combining offline metrics (AUC, NDCG) and online experiments (match acceptance, message rate).
- Build model monitoring and retraining loops informed by live interaction feedback.
WHY THIS ROLE
This is an opportunity to define the technical DNA of a consumer AI product from day one — to architect and deploy systems that combine data science, human psychology, and generative AI. Your work will directly shape how people connect, communicate, and build relationships in an AI-assisted world.