39 skills · 13 tasks · 2 jobs
Click a skill to see where it was used — the count is how many items back it.
Designed a scalable LLM-as-a-Service framework for serving large language models in a private cloud.
Delivered end-to-end LLM solutions: a RAG-based Q&A chatbot over an internal knowledge base and automated video summarization with chapter generation.
Led MLOps design on Kubernetes; implemented Jenkins CI/CD that cut ML development-to-deployment cycles from weeks to minutes.
Technical lead for the company-wide MLOps platform — built GPU services on Kubernetes to accelerate ML training/inference; integrated Apache Spark for large-scale data and ML workloads.
Led MLOps design on Kubernetes; implemented Jenkins CI/CD that cut ML development-to-deployment cycles from weeks to minutes.
Technical lead for the company-wide MLOps platform — built GPU services on Kubernetes to accelerate ML training/inference; integrated Apache Spark for large-scale data and ML workloads.
Built real-time monitoring for enterprise data jobs; optimized job-status updates from 1 hour to 2 minutes, scaling event delivery with Kafka.
Delivered end-to-end LLM solutions: a RAG-based Q&A chatbot over an internal knowledge base and automated video summarization with chapter generation.
Shipped self-serve cloud IDEs (JupyterLab, VS Code, RStudio), a no-code/low-code ML framework, and an ML-as-a-Service marketplace.
Technical lead for the company-wide MLOps platform — built GPU services on Kubernetes to accelerate ML training/inference; integrated Apache Spark for large-scale data and ML workloads.
Delivered end-to-end LLM solutions: a RAG-based Q&A chatbot over an internal knowledge base and automated video summarization with chapter generation.
Shipped self-serve cloud IDEs (JupyterLab, VS Code, RStudio), a no-code/low-code ML framework, and an ML-as-a-Service marketplace.
Open-source library for running language models entirely in the browser by compiling C/C++ inference engines to WebAssembly with Emscripten.