Generative Ai United States

Aaron Rivkin, PhD

United States Citizen ***********@*****.***

Summary

I am at the cutting edge with AI (NLP and LLM) technology with prior researcher experience, with a performance focus and strong leadership experiencel. Are you looking for Strategic and Implementation help with GenAI? Experience

Director, Logic and Code Models

OpenAI Inc

Mar 2019 - Present (4 years 1 months)

Promoted to Director of "Code" foundational LLM product and model training in Jan 2022 - Took over the software code development Generative-AI/LLM team and doubled its size to 54, restructuring into teams and moved the group from a Research mindset to a Commercial/Production one. Resulted in improved LLM reliability and consistency in delivering code suggestions (RQ>80%). Extensive customer facing work to solve their GenAI goals (Technical Product Management) and drive the team to deliver both core product and custom/consultatative work for Enterprise customers..

"Media" Foundational LLM training - Lead a team of 44 to train LLMs on social and educational media content Generative-AI, including pre-screening data for ethical content and compliance with company policies with the goal of being able to learn from existing content and create new, unique content. RQ >70% accomplished. Developed many of the core metrics used today for LLM quality control, and introduced methods to automate LLM comparative metrics. Director of AI (Contract)

Pearson

Jun 2017 - Feb 2019 (1 year 9 months)

Created the Generative AI department at Pearson in order to explore custom foundational NLP/LLMs in the creation of "addictive" content for education, analysis of essays and other student input and much more. Built the research team to develop automation of content generation and customized content based on individual learning patterns, content interests and "Fun". Headed the Ethics Review Sub- Committee.

Senior HPC Engineer, NLP/LLM Algorithms

NVIDIA

Jan 2015 - Jun 2017 (3 years 6 months)

Applied NLP algorithms to Nvidia GPU technology, collaborating with product teams to improve focused specifications, testing and overall architectures in future GPU hardware. Lead the SDK team for NLP algorithms development and support. Team Lead and LLM/NLP Researcher

MIT-IBM Watson AI Lab

Jan 2008 - Jan 2014 (6 years 1 month)

Lead a team of researchers in computational efficiency improvements in software, systems and hardware in the development of Natural Language Processing and Language Models based on Deep Neural Networks. Developed a novel way to improve neural network performance of Transformer based (BERT and early LLMs) models. Development much of what has become core technology in the training of LLMs and used in LangChain.

Postdoctoral Researcher

Massachusetts Institute of Technology

Jan 2007 - Jun 2008 (1 year 6 months)

Explored NLP (Eary forms of RNN, LSTM, Transformer Models) for analysis of content used in education for bias, tone, content accuracy and plagiarism. Education

University of California, Berkeley

Doctor of Philosophy - PhD, Artificial Intelligence Sep 2002 - Aug 2007

University of California, Berkeley

Master of Science - MS, Data Science and Machine Learning Sep 2000 - May 2002

Continuing my education in Data Science, Machine Learning and AI, started to focus on NLP, BERT models.

University of California, Berkeley

Bachelor of Science - BS, Mathematics and Computer Science Sep 1996 - May 2000

Focus on Data Science, Machine Learning and AI.

US Navy Veteran

Skills

AI Pipeline Development using LangChain and NeMo • High Performance AI Inference Pipelines • RAG

(VDB, GDB) driven GenAI • Mathematically Correct Multi-Stage GenAI • PySpark • Machine Learning • BERT, NER etc. (Language Models) • GPU • CUDA • HPC • Computer Hardware • Data Science • Python

(Programming Language) • C++ • Java • Rust • NodeJS • ReactJS • Flutter/Dart

Contact this candidate