Jeff Mo Email & Phone Number

San Francisco, Ca, Us

San Francisco, California, Us

Recommendations and Search ML

Mountain View, Wy, Us

Strategic and tactical guidance on applying LLMs in the professional services domain

San Francisco, Ca, Us

Tech leading neural retrieval and NLP. Shipped ~10 projects driving an A/B tested ~27% lift in product orders in SearchBuilt Faire's DL modelling platform, adopted by 4 workstreamsMoved search retrieval to an end-to-end fine-tuned personalised multi-modal LLM-based 2-tower model, combining engagement, text, visual and graph neural network signals. (this is not a keyword dump :))

San Francisco, Ca, Us

NLP, Representation Learning, Recommendations and Ranking

San Francisco, Ca, Us

Entity Linking, Search, Knowledge Graphs

Toronto, Ontario, Ca

Applied ML in quantitative finance and trading

Te Aro, Wellington, Nz

Developing novel deep learning models for document extraction and categorisation, and bringing them to production through Kubernetes and AWS- Leading and owning the research, architecture, and implementation on an API for grouping receipts by their contents, using multi-choice QA models based on GRUs+co-attention. The deployed model is updated in real-time through event-driven incremental finetuning on feedback, and is hosted on Kubernetes - Working in a team to bring the Text Extraction Pipeline, the first deep learning project in Xero, to production with a highly scaleable and asynchronous architecture and queue-driven Kubernetes workers, capable of serving ~1 million requests per day with the help of GPU inference and Tensorflow Serving. - Constant research into improving the accuracy and end-user value of this BERT-inspired text extraction model with novel or otherwise cutting edge NLP techniques.

Te Aro, Wellington, Nz

Prototyping deep computer vision & natural language models, along with some big data engineering• Responsible for designing a novel CNN-based model for key-text extraction from smartphone-taken images, with the effect of cutting human extraction costs by ∼70% while reducing processing time from hours to seconds• Engineered a scalable data pipeline with AWS Batch and Airflow for reproducible and traceable workflows. Started the use of PySpark on EMR for data preprocessing and distributed model training• Leading research effort on account category prediction, using state-of-the-art question-answering models. Also developed a scalable live model retraining regime such that the model always give up-to-date predictions.• Currently working with engineers to take deep learning models to production through containerisation

Te Aro, Wellington, Nz

Machine learning, platform, and full stack development- Researched and developed an accurate proof-of-concept generalised document extractor- Built a payment fraud detection model which is pending productionisation- Developing web apps with React/Redux/ES6 or Typescript on the front-end, and .Net Core on the backend, built upon AWS infrastructureTechnologies involved: C#, .Net core, MVC6, AWS, Python, Numpy, Pandas, Tensorflow, Scikit-learn, Matplotlib, ES6, React, Redux

Hamilton, Waikato, Nz

Research and implementation of a platform for automated visual classification of plant, animal or fungal species. Achieved a 19% improvement in classification accuracy compared to Google's then state-of-the-art model InceptionResnetV2. - Scraped raw images from the web to construct the initial iNaturalist dataset with 20,000 classes, 1.3m images and weighing in at 800GB.- Extended InceptionResnetV2 for high resolution images by adding an additional foveal crop column in the network- Developed novel fine-tuning technique to enhance classification accuracy when given a hierarchy- Deployed web applicatiom to enable classification on-the-go from any device- Culminated in a publication at the Australasian Joint Conference in AI and a workshop poster at CVPR in 2017 Technologies involved: Tensorflow, Node.js, Python, OpenCV, NumpyThe publication can be read here: http://researchcommons.waikato.ac.nz/handle/10289/11266

Bachelor’S Degree, Computer Science (Overseas Exchange)

Bachelor Of Computing And Mathematical Sciences (Hons.)