Hire Taixing B. - AI Engineer - New York

* Built and operated production-grade RAG infrastructure using LangChain, including document ingestion, chunking, semantic ranking, and vector store orchestration, improving grounded response accuracy by +38%. * Designed LLM-to-database access layers using LangChain SQL and tool abstractions, enabling safe, structured querying over operational data and reducing manual lookup time by 65%. * Architected multi-agent orchestration frameworks with LangGraph, enabling task-specialized agents (retrieval, reasoning, validation) and increasing end-to-end task success rate by 28%. * Implemented LLM observability and evaluation pipelines with LangSmith, including regression testing, prompt/version diffing, and pairwise model evaluation, preventing 92% of quality regressions prior to production release. * Deployed low-latency LLM inference services with online tracing, evaluation hooks, and monitoring, maintaining P95 latency under 1.5s at production scale. * Built a centralized analytics and feature data platform on Snowflake to unify multi-source operational and event data for downstream AI and ML workloads. * Applied reinforcement learning techniques (PPO, DPO, GRPO) for LLM fine-tuning, leveraging offline evaluations, human and programmatic feedback signals to optimize response quality, * Designed and operated high-throughput Kafka event pipelines to support real-time ingestion, model signals, and agent execution triggers. * Developed serverless, distributed backend services using AWS Lambda to support scalable AI workflows and data processing. * Implemented resilience patterns (timeouts, retries, exponential backoff, circuit breakers) across AI and data services, significantly improving system reliability under load.

* Developed NLP and machine learning solutions for healthcare applications using TensorFlow, PyTorch, and spaCy. * Built end-to-end machine learning pipelines covering data processing, feature engineering, model training, evaluation, deployment, and monitoring. * Designed and implemented production inference services and model-serving APIs supporting real-time clinical workflows. * Developed text classification, sentiment analysis, entity extraction, and predictive modeling solutions for healthcare datasets. * Built automated evaluation and validation pipelines improving model quality, reliability, and deployment readiness. * Designed event-driven data processing architectures using Kafka and NATS supporting real-time machine learning workloads. * Built centralized analytics, logging, and monitoring pipelines using BigQuery and cloud-native infrastructure. * Collaborated with product, engineering, and clinical stakeholders to deliver production AI capabilities.

* Developed machine learning solutions for financial forecasting, sentiment analysis, and quantitative research applications. * Built predictive modeling pipelines using structured and unstructured financial datasets to generate investment and market insights. * Designed distributed feature engineering and data processing platforms using Spark, Hadoop, Kafka, and HBase. * Developed real-time data ingestion and analytics systems supporting machine learning and quantitative research workloads. * Built model evaluation, validation, and backtesting frameworks improving prediction quality and deployment readiness. * Partnered with quantitative researchers and software engineers to operationalize machine learning capabilities in production environments.