H
Harry Wang
About
Detail
California, United States
Résumé
Jobs
verified_user
0% verified
Senior Software EngineerCrestaMay 2022 - Current (4 years 5 months)Led development of AI conversation intelligence systems for real-time sales and customer support coaching using LLMs, speech recognition, and retrieval-augmented generation (RAG). Built multi-agent orchestration pipelines with LangChain and LangGraph, adding agent routing and tool-use hooks for contextual retrieval. Fine-tuned transformer models (BERT, Whisper, DistilBERT) for sentiment analysis, intent classification and topic detection on multilingual call data. Designed evaluation frameworks combining LLM-as-a-judge, human QA review, and regression scoring to assess accuracy, latency, and interpretability. Deployed and monitored models on GCP Vertex AI and GKE, achieving sub-300 ms median inference latency and 99.7% uptime across enterpr- Senior Software EngineerAmazonDec 2018 - Apr 2022 (3 years 5 months)Developed and deployed personalization and ranking models for Amazon Retail using PyTorch, XGBoost, and AWS SageMaker to optimize product recommendations and search relevance. Built end-to-end data pipelines with AWS Glue, EMR (Spark), and Redshift Spectrum, processing 100 TB+ of daily behavioral data for model training and evaluation. Designed and productionized forecasting models (ARIMA, LSTM) for demand prediction and pricing elasticity across global marketplaces. Created internal feature pipelines and metadata tracking systems in AWS DynamoDB and S3, enabling automated feature lineage tracking and version control through Step Functions and Lambda triggers. Implemented A/B testing and model evaluation pipelines integrated with CloudWatch
- Software EngineerFacebookJan 2016 - Nov 2018 (2 years 11 months)Developed large-scale ranking and recommendation pipelines for News Feed and Ads using PyTorch, Caffe2, deployed through internal FBLearner Flow and Caffe2Go platforms. Designed and optimized feature engineering pipelines with Kafka, Spark, and Airflow, processing petabyte-scale user and content data for real-time model training. Implemented embedding-based retrieval models and similarity search systems using Faiss, improving relevance and engagement on feed content by 14%. Built internal experiment dashboards in React.js, TypeScript, and GraphQL to visualize A/B test outcomes, model drift, and performance metrics. Collaborated with applied scientists to productionize deep ranking models with distributed training on GPUs, integrating featur
- AIntern Software DeveloperAunt Bertha, Inc.Sep 2015 - Dec 2015 (4 months)Built Python-based data collection and integration tools to streamline social services data ingestion. Designed SQL queries and lightweight ETL scripts to clean and merge datasets for internal analytics dashboards. Implemented logging and validation checks to ensure consistent schema and reduce data-processing errors.
Education
verified_user
0% verified
- Master of Engineering (MEng), Computer EngineeringTexas A&M UniversityAug 2013 - May 2015 (1 year 10 months)