Senior Machine Learning Engineer

I contribute to the development and productionization of Sora, OpenAI's multimodal text-to-video generation system. I work with research and platform engineering teams to build distributed training and inference pipelines for transformer and diffusion models using PyTorch and FSDP, improving GPU utilization and reducing multi-node training time.

I design preprocessing, dataset versioning, evaluation, and experiment-tracking workflows for video, image, and text data. My work includes automated quality and safety evaluation, asynchronous inference services, optimized batching, mixed-precision execution, model parallelism, Kubernetes-based ML infrastructure, observability, deployment automation, and mentoring engineers on distributed ML and MLOps practices.

I developed large-scale machine learning models for Airbnb's search ranking and personalization platform, improving marketplace relevance and the guest booking experience. I built feature engineering pipelines processing billions of search events, listing attributes, and marketplace interactions with Apache Spark, and created learning-to-rank, recommendation, and demand forecasting models.

I implemented production ML workflows covering feature generation, training, validation, deployment, continuous retraining, and monitoring. I also designed large-scale A/B experiments, developed low-latency inference APIs, partnered on distributed ETL and feature-store improvements, and mentored engineers on scalable ML systems, software development, and API design.

I developed production machine learning solutions for a clinical conversation intelligence platform, helping clinicians and patients understand medical conversations through automated extraction and summarization. I built NLP models for medical entity recognition, clinical concept extraction, dialogue classification, medication and diagnosis identification, and clinical summarization using Python, Scikit-learn, TensorFlow, PyTorch, and transformer architectures.

I created secure, low-latency FastAPI inference services and automated ML workflows for preprocessing, training, evaluation, deployment, and versioning using MLflow, Docker, Kubernetes, and CI/CD pipelines. I also developed evaluation and drift-monitoring systems, collaborated on distributed clinical-document ETL pipelines, and maintained HIPAA-conscious data security, auditability, and operational reliability.