Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

Hi, I'm Joseph Flaten, a Senior Applied AI/Research Engineer focused on making retrieval-augmented AI practical at scale. I enjoy bridging research and product by designing robust RAG architectures, embedding systems, and evaluation frameworks to improve grounding, faithfulness, and search quality. In my day-to-day, I prototype models in PyTorch and TensorFlow, build feature pipelines and in-memory data experiments, and collaborate with GenAI teams to ship dependable GenAI solutions. Outside work, I enjoy exploring new AI tooling and sharing findings with product teams.…Hi, I'm Joseph Flaten, a Senior Applied AI/Research Engineer focused on making retrieval-augmented AI practical at scale. I enjoy bridging research and product by designing robust RAG architectures, embedding systems, and evaluation frameworks to improve grounding, faithfulness, and search quality. In my day-to-day, I prototype models in PyTorch and TensorFlow, build feature pipelines and in-memory data experiments, and collaborate with GenAI teams to ship dependable GenAI solutions. Outside work, I enjoy exploring new AI tooling and sharing findings with product teams.

Joseph Flaten

AI Engineer, Data Scientist, Full Stack Developer, +2





Hi, I'm Joseph Flaten, a Senior Applied AI/Research Engineer focused on making retrieval-augmented AI practical at scale. I enjoy bridging research and product by designing robust RAG architectures, embedding systems, and evaluation frameworks to improve grounding, faithfulness, and search quality. In my day-to-day, I prototype models in PyTorch and TensorFlow, build feature pipelines and in-memory data experiments, and collaborate with GenAI teams to ship dependable GenAI solutions. Outside work, I enjoy exploring new AI tooling and sharing findings with product teams.…Hi, I'm Joseph Flaten, a Senior Applied AI/Research Engineer focused on making retrieval-augmented AI practical at scale. I enjoy bridging research and product by designing robust RAG architectures, embedding systems, and evaluation frameworks to improve grounding, faithfulness, and search quality. In my day-to-day, I prototype models in PyTorch and TensorFlow, build feature pipelines and in-memory data experiments, and collaborate with GenAI teams to ship dependable GenAI solutions. Outside work, I enjoy exploring new AI tooling and sharing findings with product teams.

Available to hire

Hi, I’m Joseph Flaten, a Senior Applied AI/Research Engineer focused on making retrieval-augmented AI practical at scale. I enjoy bridging research and product by designing robust RAG architectures, embedding systems, and evaluation frameworks to improve grounding, faithfulness, and search quality.

In my day-to-day, I prototype models in PyTorch and TensorFlow, build feature pipelines and in-memory data experiments, and collaborate with GenAI teams to ship dependable GenAI solutions. Outside work, I enjoy exploring new AI tooling and sharing findings with product teams.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Language

Javanese

Intermediate

Work Experience

Senior AppliedAI/Research Engineer at Elastic

March 1, 2021 - Present

Led applied research on Retrieval-Augmented Generation (RAG) architectures, benchmarking dense vs sparse embeddings (ELSER), multi-stage reranking, and hybrid retrieval strategies to improve LLM grounding, accuracy, and search relevance. Designed novel embedding & similarity pipelines within Elasticsearch, evaluating vector indexing, ANN search (HNSW), and retrieval fusion for multi-billion document clusters. Built document understanding and context construction frameworks for LLMs to enhance downstream generative answer quality. Conducted large-scale offline experimentation and ablation studies, measuring NDCG, MRR, recall, and hallucination rates across configurations. Collaborated with GenAI teams to prototype retrieval-augmented generations systems, integrating external transformers and OpenAI-compatible APIs for experimental evaluation. Published internal research reports to guide product deployment.

Senior AI/ML Engineer - Personalised Intelligence at Uber

October 1, 2019 - February 1, 2021

Built and productionized personalized recommendation and ranking systems for Uber’s commerce surfaces using content-based models and learned product embeddings to optimize discovery, ranking, and conversion at scale. Designed end-to-end ML pipelines on Uber’s data stack with Databricks, Airflow, and distributed feature computation to support offline training, online inference, and continuous model refresh for high-traffic user-facing services. Developed a generative AI–powered visual outfit recommendation experience using diffusion models and style embeddings, driving an 18% lift in conversion.

Machine Learning Engineer at Cognizant

August 1, 2015 - September 1, 2019

Designed and deployed an end-to-end financial predictive analytics platform for enterprise clients, transforming large-scale operational and behavioral data into real-time decision signals using Python, TensorFlow, and Scikit-Learn. Built supervised and time-series models (XGBoost, LSTM, Prophet) to forecast demand and resource utilization, enabling business units to reduce operational costs and improve planning. Engineered feature extraction and model training pipelines processing millions of records per day using Apache Spark and distributed data workflows. Implemented model versioning, experiment tracking, and automated retraining with MLflow, improving reproducibility and reducing drift in production. Deployed machine learning services as containerized REST APIs using Docker and Kubernetes, integrating with GCP and Azure for low-latency inference at scale.