Alan Hogue Email and Phone Number

Working primarily as ontologist and data architect for a product which helps institutional investors vet possible investment opportunities. It imports user data, integrates it into an ontology, and generates investment memos which are tailored to their specific investment policies and philosophy, as well as assumptions which may affect investment decisions. This position also includes a fair amount of coding, as well as leading a small team of developers and doing API design.

- Senior NLP engineer focusing on automatic generation of knowledge graphs from unstructured text- Created custom ontologies for specific industries and applications, including tech and automotive spaces- Led major design changes to a complex Java application, including integration of coreference resolution and LLMs, leading to significant improvements in accuracy.- Mentor junior NLP engineers

- Developed a new low-resource speech-to-text system suitable for use on IoT devices, allowing the Mycroft platform to move away from reliance on Google's TTS service.- Improved NLP system which allowed users to control applications with voice commands - Developed applications for the platform for controlling smart home devices, internet radio, etc.

- First employee. I was closely involved with everything mentioned below from inception and worked on nearly every part of the system.- Was a key software developer given wide responsibilities to analyze, identify, design and implement new components in a search product based primarily on ElasticSearch, Metaflow and Serverless- My goal was to create a search product with advanced semantic features, including conversational question/answer sets based on a carefully fine-tuned GPT-J model, a language model (perhaps the first of its kind) specifically adapted to understand the very strange and slang-y discourse in the crypto/NFT space, including peculiar uses of emojis.- Computational efficiency was a must: used packages such as concurrent.futures and NMSLIB to speed up computation.- Everything lived in an AWS Batch-based metaflow pipeline which ingested tweets/pages and parsed them extensively, depositing the results in an ElasticSearch instance.- Created APIs for the front end team.

Led a team in designing an ontology and associated grammars for technical support applications.

Responsible for development of text classifiers using SVM and deep neural network models. This included supervising large annotation projects, drafting category definitions and annotation guidelines, training, evaluating and debugging models, and launching models into production. Responsible for maintaining large specialized ontologies for B2B applications.Highlights:- Launch of text classifiers for new AdSense for Urdu- Development of highly sensitive Ads Policy classifiers to identify things like hate speech, content relating to drugs, and so on

Worked on large scale projects integrating WordNet and other ontologies into the Knowledge Graph.

Designed and developed a national database of the federal judiciary, including a web front end for use by law students seeking employment therein. Managed several existing databases, including full responsibility for data integrity and for providing assistance to end users as needed. Provided consultation regarding the acquisition of new technology assets, planning for future database needs, etc.