Sign In

Michael O'Connell

Michael O'Connell is a prominent figure in the field of analytics, holding the position of Chief Analytics Officer at TIBCO.

He pursued his academic career with a Doctorate in Philosophy in Statistics from North Carolina State University and a Master's in Statistics from UNSW.

Throughout his professional journey, he has held key roles in various organizations, including Chief Data Scientist, Senior Director of Analytics, and Vice President Life Sciences at TIBCO Software Inc.

Additionally, Michael has served as the Vice President of Software Products at Health Hero Network, President at Waratah Corp, showcasing his versatile expertise across multiple industries.

Highlights

Jul 9 · linkedin

SEMICON WEST is abuzz this year.. breakout of Gen-AI!!

Kickoff with Marco Chisari, EVP US Samsung Semiconductor and Innovation Center. Marco threw out a challenge - can we develop a new compute architecture vs CNN multiplication? As it stands, current growth in AI data center power consumption, will use all available world energy production by 2030.

As context, there is a major increase in demand for DRAM, with advent of high bandwidth memory (HBM). Samsung is the biggest DRAM producer by revenue, and offers HBM customization based on workloads for optimization.

The thread on power consumption and memory with AI continued with super presentations from Subhasish Mitra, Stanford, Ameen Akel, Micron and John Hu, NVIDIA in the Future of Computing breakout session.

Mitra showed cutting edge work on new compute paradigms like Dendritic algorithms .. more than 3 orders of magnitude lower energy consumption than CNNs according to recent Nature article.

https://lnkd.in/eUsVs-Zt

Akel showed practical examples of compute limitations with current models and need for optimization of AI workloads. How to increase bandwidth while keeping energy constant. Akel posited that we need a 324X improvement in memory and or interconnect efficiencies in order to meet reasonable response times in current-next-gen Gen-AI use cases.

Logic is scaling well; DRAM needs to scale more for constant energy; Data movement needs to scale; Advanced packaging is an oppty.

Hu added an NVIDIA perspective .. many products and flashy videos :-)

From there to CXO Panel on AI/GenAI with Dr. Bobby Mitra moderating — very innovative use of GenAI at Samsung with MMFM for Etch, Yield, RCA, Recipe Optimization presented by Jae-Yong Park. Fascinating use of LLMs for time series forecasting..

I am thrilled for Spotfire to be part of this HTM community, especially in the DRAM HBM area. It feels like memory could become center of entire compute stack..

Our work on wafer defect analysis, anomaly detection and RCA; and our multimodal Spotfire Copilot is on point. Fun to meet with Spotfire customers and hang out with Evie Klaassen, Spotfire HTM Data Scientist.

https://lnkd.in/eSeVcE5h

https://lnkd.in/ecWq3PSH

Spotfire #VisualDataScience

Marco Chisari Ameen Akel Subhasish Mitra John Hu Pradip Singh Daniel Gamota Surya Iyer Janhavi Giri, PhD

SEMICON WEST is abuzz this year.. breakout of Gen-AI!! 

Kickoff with Marco Chisari, EVP US Samsung
Michael O'Connell
Add to my network

Location

San Francisco Bay Area