Groq’s Language Processing Unit™ (LPU), an AI inference accelerator, and Cloud Platform have set the bar for fast inference of generative AI. Groq offers inference on large language models (LLMs), speech-to-text, and vision models that run our custom software stack. Directed a high-performance software engineering team that developed and optimized cutting-edge tools, frameworks, and APIs, ensuring seamless integration and maximizing the capabilities of Groq's LPU. Delivered transformative results, including significant improvements in throughput, latency, and energy efficiency across AI workloads. Led efforts to enable algorithm design and fast deployment. Results enabled $640M funding round in 2024.

• Provided technical and strategic guidance to clients in machine learning and functional programming.• Advised family offices on tech investment decisions.• Collaborated with teams at Tweag.io and Bain & Company to develop innovative solutions for clients' needs.• Achieved measurable impacts on revenue and efficiency through strategic recommendations.

External Advisor to Bain & Company's amazing Advanced Analytics Group (AAG). AAG is helping Bain's clients deliver value from data, analytics and AI to the bottom line.

Technical and Business Development at one of the most remarkable tech labs in the world, tweag.io.

Built and led team of data scientists, data engineers, and big data platform specialists for the best loved retailer in the U.S. Joined as Target reorganized for a successful digital transformation. Grew a global AI team by over 4x and expanded scope of efforts to many parts of Target, including personalization, targeted advertising, enterprise forecasting and more. Introduced deep learning, product management and agile methods to the standard practice of data science at Target.

Launched mobile app called Grocery Exchange to empower shoppers with a platform to share and compare grocery prices. This advanced mobile shopping list will enable families to save billions on their grocery bill. Crowdsourcing data has proven to be an effective and inexpensive means for gathering data in apps such as Waze and GasBuddy. Pingwell supports millions of user sessions a month. Designed and built the computer vision system at the core of the key feature of Grocery Exchange--the ability to extract data from non-standard shelf labels. Hosted on the cloud, our data pipeline extracts information from shelf labels and shares it with other shoppers. Built an expert team of app developers and led design of new mobile app to provide this new capability to our users.Developed and executed brand marketing strategy that allows Pingwell to sell native ad space at many multiples of generic ad engines. Grocery Exchange can offer grocery brands a 1-to-1 channel to reach shoppers before they shop, or while they are in the store. Designed a gamification system that incentivizes shoppers to collect data with ease.

Renewable energy company focused on increasing the supply of sustainably produced biodiesel and green jet fuel. Biogreen Oil garnered a $150mm investment from a major international pension fund. Developed investment strategies and protocols, set up local operating companies, evaluated local partners and governments and conducted project assessments in over a dozen countries. Built a high resolution global project suitability model based on latest global climate models and land use data from satellite imaging. We combined this with custom project reporting and video sharing tools using Google Earth. Managed external communications and investor relations and lobbied the Australian and Indonesian governments for regulatory changes for our industry.

Morgan Stanley is the place that sparked in me a deep interest in technology and better ways to build software. Banking and finance provide some of the ultimate tests of systems made by people and of people. After 8 1/2 years at MS and 20 years in business I headed off to earn a PhD in machine learning.Before that I worked in Fixed Income Technology Management. Responsibilities included forming strategy for 300 programmers in the Fixed Income Division (FID) technology team. We designed, developed systems, and managed teams building systems for trading and sales platforms across the division. FID is responsible for trading bonds, currencies and commodities. For example, we designed a new issue trading platform for worldwide use across products. I also designed and deployed an intranet application that provided internal trading color, live pricing information, client trading flow, and “heat-mapped” relative value information to 900 traders and sales people worldwide. This innovative system was a "Web 2.0" application well before the term was widely used, and was the first such division-wide information sharing application at Morgan Stanley.Prior to that I worked in the Securitized Products Group (SPG). I built and managed risk reporting for a $15 billion balance sheet. I instituted risk controls and regulatory capital management techniques, and evaluated trading strategies and security pricing models.I was originally hired to do product management for the Mortgage Department. I designed mortgage trading systems, securities valuation models, and managed business development for our IT staff. I was responsible for marketing a wide array of mortgage and asset-backed securities.

Proud to have worked at a great firm with talented colleagues who challenged me to contribute in so many different ways.My first assignment was to work on Market Research. I led the market research group for mortgage securities, publishing over 200 weekly reports plus numerous white papers. I designed and implemented the first mortgage valuation system at Goldman and incorporated it into firm’s research and trading systems. The integration required by these models is still impressive, combining prices from thousands of traded products from across the firm and over a dozen econometric models for prepayments and spread relationships. To outsiders these models were black boxes. However they were designed to be intelligible to traders and risk managers. Other highlights include the first implementation of the Black, Derman,Toy interest rate model for the mortgage market on the Street. I also discovered a variance reduction technique that sped up calculations by a factor of 10. Back then we rented time on a Cray supercomputer for about $10,000 a month so this translated directly to large dollar savings. I was seconded for a year to help form the new Goldman Sachs Asset Management (GSAM) division, The fixed income portfolio started with some corporate bond assets and a global mutual fund that I helped raise internationally and then sole-managed. This project Inaugurated the long-term fixed income portfolio for GSAM, which has since grown to nearly a trillion dollars of assets under management. In London, I worked on European Bond Markets. There I researched a wide range of products including German, Dutch and Austrian government bonds, DM futures markets, perpetual floating rate notes, ECU-denominated bonds, credit spread options, dual-currency and partly collateralized Eurobonds. Our group launched the publication of the weekly European Bond Market Commentary and we celebrated in a 400 year old pub after sending it to the printer each week.

Joined Shearson Lehman Brothers straight out of grad school. Anybody remember Shearson? EF Hutton? Extra points for Mabon Nugent, Manny Hanny, or Keith Hernandez.Hired into the hot product area of the day, Mortgage-Backed Securities (MBS). Back then the credit quality of MBS was so high, defaults were just an afterthought. The market was growing fast as falling oil prices, lower interest rates and a booming economy made the demand for mortgage credit very strong. Rates and spreads were very volatile and the primary mortgage lenders of the day were bad at managing risk and essentially disappeared from the markets. Investors urgently needed help understanding all the new products hitting the institutional market.My first assignments as a freshly minted quant were clear. 1) Copy statistics by hand from large reports into smaller reports. 2) Walk data tapes late at night from one data center to another. I was pretty good at walking, but my handwriting is bad. This caused problems, so even though I wasn't hired to write code, I wrote a program to eliminate my job.Dues paid, I was soon assigned to write research reports and new product white papers. We also developed new products and I wrote the first Street publication on interest-only and principal-only mortgage-backed securities. Designed and implemented Lehman's first mortgage valuation model for adjustable rate mortgage securities.