Leading a team for Multimodal LLM infra. WE ARE HIRING strong ML engineers who have direct experience in LLM and multimodal LLM. If you are interested, please send your resume to recruitment@anuttacon.com. Several matching key words:LLM & Diffusion training and serving, NCCL, Customized PyTorch, Profiling, Quantization, Distillation, Large Scale data processing (Spark, Airflow, or equivalent) , Gang Scheduling, Fault tolerance, GPU health.

Cofounder of Fuyao Super Computing Center (https://cnevpost.com/2022/08/02/xpeng-builds-largest-self-driving-computing-center-in-china/). R&D head of Infra and AI platform.Leading a cross-company effort with hundreds of engineers to build an Exaflops hybrid cloud computing infrastructure including hardware & software stack. Advancing A.I. productionization.#HPC #distributed storage #distributed training&inferencing #Containerization #Orchestration and automation #Virtualization

● Led Inceptio's engineering effort in autonomous driving as the head of US infrastructure as well as managing the ML infra team, Data platform team, Frontend team and DevOps team. (40+ hc)

● Joined as the founding engineer of ML platform team. Grew the team from scratch. Established multiple company-wised standards and business processes for data and machine learning, and built multiple core systems to distill knowledge from massive sensor data streams.● Established ML model generation, release and triage process. Built a k8s based machine learning platform which maximizes GPU cluster usage, incubates (AutoML), automates evaluation and scales model serving. The platform also manages model CI/CD, regression test and integration test.● Established data labeling and triage process. Unified and established company-wised annotation standard.● Built a large scale data warehouse system that serves end-users with human annotation, machine-in-the-loop auto annotation, data transformation, semantic search, data mining, statistics and visualization.

Constructed end-2-end machine learning solution. Multiple launches continuously improves recalls and coverages which help Google Assistant to better understand human intents.

Initiated, lead and managed the Gifter product. Rallied a team of 5 googlers and participated in Area 120 (Google internal incubator) on 2017.

Worked in Shopping Search Quality team. Drove, coordinated and developed multiple machine learning based projects across teams for better shopping search ranking experience. Multiple launches cumulatively contributed over $120M revenue grow.

CSE535 -- Information RetrievalEAS230 -- Computing and Numerical Methods for Engineers

Interned as Deep Learning Scientist. 1) Designed session-related features from billions of in-game mobile device event logs. Perform A/B tests and improved the AUC scores on conversion rate prediction for over 70% (1k+ contracted) ad-offers. 2) Initiate and implement Deep Learning pipelines from terabytes of data for device level sequential behavior modeling. 3) Design, implement and optimize experimental pipelines with Hadoop Streaming on Amazon Elastic MapReduce framework. Cut 80% of the processing time from former approaches