
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
ML/AI Founder | Focus on: LLMs, AI+code, recommendations
I'm an AI / machine learning founder, engineer, and manager. I have a history of either starting companies (ex-YC), or joining small teams, such as being an early engineer at Pulse (now part of Linkedin), Medium, and Primer AI. My PhD and career have both revolved around data-centric and AI-based systems. For the past 6 years, I've managed a team of ML engineers working as technical consultants and contributors for dozens of different clients. With those clients, we've built a wide variety of custom AI-based tools that have been integrated into production-facing sites and apps. I'm passionate about building technically innovative products and bringing together amazing teams.
New York University
PhD, Applied math
January 1, 2001 – January 1, 2006
Unbox Research
Machine Learning Engineer / Founder
October 1, 2017 – Present
San Francisco Bay Area
Primer AI
Machine Learning Engineer
September 1, 2016 – October 1, 2017
San Francisco Bay Area
OpenAI
Machine Learning Engineer (Contractor)
June 1, 2016 – July 1, 2016
San Francisco Bay Area
Medium.com
Data Scientist / Consultant
May 1, 2013 – December 1, 2013
San Francisco Bay Area
AppGrok
Founder
April 1, 2012 – April 1, 2013
Zillabyte
Cofounder, CTO
August 1, 2011 – March 1, 2012
Mountain View, California
Storm8
iPhone app developer
January 1, 2009 – January 1, 2010
Bynomial, Inc.
Owner-employee
January 1, 2009 – January 1, 2011
Software engineer
August 1, 2006 – December 1, 2008
Software Engineer
January 1, 2006 – January 1, 2008
Created and Taught an LLM Workshop
June 1, 2023 – July 1, 2023
In July 2023, I co-designed and co-taught a full-day workshop on the engineering aspects of hosting an LLM for business applications. This was a deep dive on the technical side of things, beginning with an understanding of the math and foundational theory behind LLMs, going over retrieval-augmented generation, and covering several practical techniques for efficiently fine-tuning your own model, such as low-rank adaptation and quantization. The workshop format was an alternation between live lectures and Jupyter-based interactive sessions. We spun up one GPU instance per participant, and prepared llama-2 on each instance, as well as ensuring that CUDA was properly configured and that we were able to fine-tune using a handful of relevant Python libraries. I collaborated with Gabriel Bianconi to design and teach this class.
Created and Run the Learn & Burn AI Newsletter
February 1, 2023 – Present
In 2023, I put together a small team to create a high quality weekly newsletter aimed at AI engineers and enthusiasts. Each week, we choose a new AI research paper and explain in easy-to-read yet technical detail what the paper contributes, and how the authors achieved their contribution. What makes the newsletter great is the unique quality of the articles. Each article is much more concise than the underlying paper, and is written in a casual tone. At the same time, my team works hard to avoid oversimplification, or leaving out key details. This combination of clarity and high technical detail is extraordinarily challenging, and I'm grateful to have such an amazing team to pull this off. Each article is edited by an ex-O'Reilly editor (Dawn Schanafelt) to help us achieve a consistent, casual, and easy-to-understand tone. Our primary writer, Adrian Wilkins-Caruana, recently received his machine learning PhD, specializing in applications of AI to the medical industry. Our illustrator, Giulia Zerbini, is experienced in the design of data-oriented visualizations and product interfaces. Other staff members help to promote articles, and day-to-day operations are coordinated by Heather Hambley.
Nexus: SaaS software for personalized recommendations
January 1, 2023 – December 1, 2023
In 2023 I built a system to provide high-quality personalized recommendations via an on-premise API for use by client companies. The system integrated a number of different machine learning-based signals of recommendation quality, and was designed for use by product-minded domain experts, rather than for use by engineers. In addition to featuring a full-control browser-based GUI, the product included these features: * Extremely fast queries: API calls returned in milliseconds, designed for production use. * Support for custom quality signals as well as customized business logic for the final recommendations. * Support for ML-based training on data sets too large to fit into memory. * Tools to enable quick search and visibility of recommendation quality. * Tools to enable different forms of feedback on recommendation quality, to be used as training data. Nexus was built primarily with Python (for the engine and back-end) and JavaScript (front-end), while the systems involved with machine learning training used libraries built with C, C++, and Fortran for greater efficiency.
Cultural Fit Analysis
The candidate's diverse experience, ranging from founding startups to working at major tech companies like Google and OpenAI, indicates adaptability and a broad understanding of different organizational cultures. Their involvement in personal projects like the AI newsletter and LLM workshop demonstrates a passion for the field and a commitment to continuous learning and contribution, which aligns well with innovative and growth-oriented cultures. The focus on building user-friendly ML products (Nexus) and explaining complex topics clearly (newsletter, workshop) suggests a collaborative and impact-driven mindset.
Soft Skills & Operational Fit
The candidate's entrepreneurial background and experience in leading teams suggest strong leadership, problem-solving, and project management skills. The creation of a technical newsletter and LLM workshop indicates a proactive approach to knowledge sharing and community building. The description of the Nexus project highlights a focus on user experience and product-minded solutions, suggesting good operational fit for roles requiring both technical depth and business acumen.