Currently, I am a Senior NLP Scientist, working on Quark's proprietary customer-service management technology.

This consultancy focusses on helping businesses with significant AI-ML operations to create client-focussed solutions internationally. I divide my time between Latin America and the US. Projects include:* Creating a hunger relief algorithm for the Brazilian government, sponsored by the Gates Foundation.* Statistical inventory management for several Brazilian clients.* Use of Machine Learning and Decision Trees to manage hypothesis generation directly from the data.* Transformer-driven fine-tuning of GPT-Neo and T5 transformers for automated content generation for various European companies.* GPT-driven headlines prediction for major Brazilian news outlet.* Customized search engines and database views for legistlative and bureauocratic documents in various languages.

This company is funded by the Stanford Research Institute, which has licensed several cutting-edge NLP technologies for disease diagnosis from patient-doctor Natural Language text. Working from the US and Brazil, my responsibility is to find ways to leverage Stanford's IP in a production environment - deriving coherent semantic, first-order predicate logic expressions to perform true logical inference - and to use proprietary Natural Language Generation techniques to present clinical summaries. I am also responsible for our digital marketing campaign.

This Brazilian-American fintech company with offices in Dallas, San Francisco, and São Paulo specializes in a conversational assistant app to organize personal finances and to make trading recommendations. Working in both the US and Brazil, I am responsible for leading the AI team, developing, testing, and rolling out the product.

Nama is a Google-accelerated company building AI conversational interfaces for Brazilian, Latin American, and North American markets. Our clients span multiple industries, including government and finance. I am responsible for AI research, including entity extraction, intent determination, and logical induction.* Have done multiple presentations to Google, American investors, and the Brazilian Government. After one of these group presentations (involving the whole team), we received a $700k USD investment in the room.* Helped create promotional videos and AI get-togethers.* Participated in the Google Entrepreneurs Exchange in Waterloo, Canada.* Invited to participate in the Creative Destruction Labs program, in Toronto, Canada.

This global media and technology company incorporates and innovates the latest Machine Learning Technology to deliver to clients.* Analyze and develop Sentiment Analysis algorithms.* Development in C/C++ and Python.* Bloomberg sentiment data feed, which I was responsible for, generated $1.2m in billings.* Also responsible for the Bloomberg Social Velocity terminal function, which received widespread media attention and 4k hits a day.

This defense contractor services government clients with a wide range of capabilities, including signals processing technology, Big Data Analytics, systems integration, and intelligence analysis. In the past two years, the Washington Business Journal has ranked us as the 4th and 10th fastest growing companies in the D.C. area. While working here, I applied for a patent for a search engine, which was recently awarded (#9,176,949). I also won a government grant to implement this search engine.* Acted as project manager for a team committed to applying Cloud-based solutions to large-scale Open Source data mining technologies.* Acted as project manager for two other grant-related projects involving machine learning applied to network intrusions and security of cyber-physical systems. * Designed patented search engine algorithm targeting semantic relationships in natural language with eigenanalysis.Implemented this algorithm in MapReduce, completed in three months. Project involved: - Processing raw text to extract semantic dependencies. - Performing mathematical analysis in a distributed, parallel framework. - Indexing results into a database. - Rapid vector-space search algorithm which runs in less than O(n) time for n vectors. - Interactive GUI to submit queries and view results.* Helped synthesize output of multiple text analytics utilizing statistical parsing.* Researched applications of proprietary machine-learning Latent Semantic Indexing technology, and compared to Open-Source technology.* Successfully presented company technologies to various government clients.* Researched and presented applications of adiabatic quantum computing.* Innovated new methods of distributed eigenanalysis.* Won the Invertix Eagle Award for Technological Innovation.

My unique mix of technical and linguistic expertise allows me to provide rapid solutions to clients in the technology sector and the defense/intelligence community. * Refined an Optical Character Recognition algorithm for Arabic and Chinese.* Designed an algorithm that selected a text from a corpus, with text used to acquire handwriting samples in the Middle East.* Researched and wrote reports on various cultural, technological, and linguistic issues, including Arabic entity extraction, the Arabic script, the Chinese script, and benefits of different learning algorithms.* Performed voiceovers in Arabic, English, and Spanish.* Edited and proof-read translations in various languages, including Spanish and Arabic.* Taught classes in software development, English as a second language, Modern Standard Arabic, Egyptian Arabic, and Levantine Arabic.

This cybersecurity firm protects clients against the numerous threats to their infrastructure, intellectual property, and employees with technologies that identify malware, phishing scams, social network attacks, credit card fraud, and identity theft. Servicing private and government clients such as AOL and Microsoft, they were recently acquired by the defense contractor QinetiQ. * Designed and implemented an information retrieval system, using a competitive learning algorithm.* Designed and implemented algorithms for recognition of malware, malicious code, and obfuscated code, using language models, behavioral analysis, and HTML hierarchies.* Evaluated and adapted statistical learning software for purchase.* Ported a high-efficiency HTML parser.* Performed localization for Asian and Middle Eastern languages.* Analyzed intelligence in Arabic, Farsi, Turkish, and Urdu.

Taught programming in Visual Basic.

This laboratory at the University of Maryland focuses on innovating educational technologies that strengthen the language capabilities of government clients. Focusing on Middle Eastern, Central Asian, and East Asian languages, they are one of the main innovators in integrating technology with multilingual intelligence analysis. * Produced preliminary designs for Langnet, an interactive online curriculum for less commonly taught languages.* Reviewed documents in Mandarin, Turkic languages, Arabic, and Urdu.* Researched cognitive models for language pedagogy and languages such as Chechen and Panjabi.* Designed and implemented an Urdu Nastaliq font and an algorithm to display the font, using numerical properties of Unicode.