As the product lead for a Google OKR to drive Research into Production, I partner with Google Deepmind, Google Research and Google Product teams (Ads, Search, YouTube, Cloud, Bard, Workspace, Photos, etc) to deliver state-of-the-art innovation from research to production and build developer tools to increase velocity, model quality and safety across diverse use cases and surfaces.

Leading a team of 20+ PMs, partnering with 500+ engineers, to democratize state of the art ML for the worldCloud TPUs, On Device ML, Generative AI, Applied ML, ML Experimentation and Workflow Automation solutions

[Candidate Labs]I worked with the Head of Data Science and CEO to design the machine learning product architecture for recruiter enablement, prioritize data projects and define scope and specifications, and introduce and enforce best practices for validating ML models and deploying models in production. In a 7-week engagement, I helped the team achieve improved alignment and clarity on the prioritization of work, demonstrated how to scope machine learning research and product development effectively upfront for both supervised and unsupervised models, introduced processes for working with Engineering and Product teams to ensure stakeholder buy-in, led two models end-to-end from specification to validation, and provided the executive team with an assessment of opportunities and challenges.[The Cusp]Advisor for menopause prediction experiment design, data collection and modeling

After hiring Dan Vradenburg as CEO, I transitioned to focus my responsibilities on internal operations, scientific and technical strategy and execution- Oversight and management of production testing operations and automation to scale up testing and improve turnaround time- I led the IP assessment and strategic acquisition of the Solum facility, the technology and lease transfer, hiring of the core team, and the spin up of the laboratory- I drove strategic technological partnerships and relationships, including execution of outsourcing, automation, and build-or-buy strategies- I led product management for Microbiome and AI products, including development of new soil microbiome indicators and development of soil syndrome models to improve soil microbiome interpretation- I developed and oversaw execution of the Scientific and Technological strategy, including AgAI and the Soil Intelligence Engine- I led the data strategy around key commercial partnerships, external data integration, and digitization of agronomic insights- I led the IP strategy roadmap and development of the IP portfolio

Trace Genomics has developed the first analytics engine that learns as it maps the living soil. We are an interdisciplinary team of bioinformaticians, data scientists, software engineers, and molecular biologists, working together to serve the farming community and help growers maximize the value of every acre. As Cofounder and CEO:- I raised 22M in financing and grew the company to 45 employees- Led the development and launch of the "23andMe for Soil" test for soil microbials- Led the design and implementation of the machine learning and genomic analytics platform- Developed the GTM strategy and product differentiation for the test and the analytics platform- Recruited engineers, data scientists, product managers, soil scientists, sales associates, and agronomists at the management and IC level- Direct reports: engineering, data science, product (2015-2019), and lab operations (2019-)

Joined company as employee #10 while incubating at Khosla Ventures. Took on product management, business development, and backend engineering roles. - Led product development of medical imaging product, user story mapping and interviews with several hospitals- Supported CEO in customer contract negotiations and helped to win critical contract with Virtual Radiology group. - Led the development of customer-facing predictive solutions on the MetaMind platform, including intracranial hemorrhage and diabetic retinopathy- Implemented deep learning algorithms (Python) for natural language processing to predict entity based sentiment.

Developed predictive analytics solutions for enterprise clients across a range of verticals, including:- geotemporal crime prediction- customer analytics product for small merchants- health care fraud prediction and triage- merchandising prediction for a large Japanese convenience chainWorked with clients in US and Asia to turn new opportunities into multi-year contracts.

Developed business analytics reports for leadership, gaining insight into customer retention and growth metricsDeveloped log analysis scripts for engineering to understand bug reports and triage sources of product failure

Built a recipe search app that utilizes graph analysis to organize recipes by ingredient(Slides: http://www.slideshare.net/dna_nerd/diane-wu-insight-final-demo)• Clustered over 100,000 recipes based on the Jaccard distance of ingredients (calculated using MapReduce on Amazon EMR), labeled clusters using ngram analysis, and identified ingredient differences using log-2-ratios of vector representation• Used Python (numpy, scipy, nltk, networkx, scikit-learn, BeautifulSoup, mrjob), MySQL, MongoDB, HTML/CSS/JS/Flask

See [[ Projects ]] section for more detailed descriptions.Main Thesis Topic (Sept 2009—Sept 2011): Elucidation of RNA editing function using bioinformatic analysis of mRNA and small RNA populationsSub Thesis Topic (Aug 2010—Dec 2012): Modeling and analysis of time-series gene expression data during heat-shock response and recoverySide project (Nov 2011—Dec 2012): Applications of sequencing and metagenomics for molecular detection of pathogens in the environmentSupervisor: Dr. Andrew Fire

Science demonstrations for visitors of the museum

High-throughput sequence analysis of molecular signatures from clinical samplesIncorporating CNV, cDNA microarray, and genomic structure data to identify causative factors for inherited diseaseDeveloping algorithms for detection of miRNA editingSupervisor: Dr. Marco Marra

Validation of novel predicted genes in Pseudomonas aeruginosaSupervisor: Dr. Fiona Brinkman

Modeling techniques for improvement of transcription factor binding site predictionSupervisor: Dr. Wyeth Wasserman

Improved computational methods for orthologous gene prediction.Supervisor: Dr. Fiona Brinkman