Experience bests YouTube
Data science means serious money, plenty of information put forth online is actually selling you technology, solutions and expertise. Without a keen eye for a fit both at the problem and results level, it is easy to fall prey on claims that do not translate to project success. Dr. Duboue reviews about 40 papers a year in major AI conferences and has an expert eye to tell such things apart. And he runs his own business.
Technology that serves business goals
Often times practitioners choose technology for personal rather than business reasons. However, complex technology fresh from research is often not field tested, contains many bugs with unclear maintenance, plus it might be encumbered by patents. Dr. Duboue has 22 years of experience in the field which makes him uniquely suitable to try both new and traditional approaches to find the best solution for your problem.
Avoid unnecessary R&D risks
Many problems can be solved by better data cleaning and other, somewhat drudgery processes. Or with yet to be invented algorithms. The excitement of a possible scientific discovery might derail your data science efforts. Dr. Duboue years of corporate R&D experience helps you avoid such risks before they take place.
Foster company-specific solutions
For an effective investment, your in-house talent needs to expand and ground their technical knowledge with insights and feedback from the rest of the company. Dr. Duboue can help mediate this process, highlighting the value to each party in the collaboration by tapping on his experience of more than a decade on this process.
Develop your human resources
Mentoring from an external expert with an established career and reputation can be life-changing for your team. Dr. Duboue's has more than 50 papers and patents published over the last 20 years that have received close to 2,000 citations and a machine learning book published in 2020 by Cambridge University Press.
Transform your codebase
Dr. Duboue is an unusual AI expert who is still an active coder with 40+ projects on GitHub. He can provide you with actionable transformations on your codebase, including technology migrations and (re-)architure.
The courage to say “I don't know”
Data science is a new field and there are many efforts that might not pay off. Dr. Duboue only accepts a project if assessed to recoup investment. Reach out to see whether he might be a good fit to your needs.
What can Dr. Duboue do to help your business?
Custom machine learning metrics based on the unique characteristics of your business: ML optimizes what you ask to optimize, using off-the-shelf metrics might leave very expensive errors unnoticed.
Assessing existing human and computational resources: for hiring, retaining, purchasing and maintenance.
Architecting a complex data science solution with a focus on long-term maintainability: one shot vs long shot.
Communicating existing solutions to the business side of your company to enable effective feedback: Dr. Duboue runs his own business and he presents data science topics in business terms.
Type of contributions done for other businesses in the past:
- Large Language Models: experience fine-tuning models, evaluating them, designing AI products on top of them, and assembling training instructions, both with open source models and OpenAI API (ChatGPT, GPT4).
- Chatbots: ample experience in the field, including designing dialogue trees, training models from scratch and model maintenance.
- Semantic search, customized IR: enhancing search based on business specific annotations.
- Custom text analytics pipelines: automatic annotation of large volumes of text based on business-specific types of information, including writing annotation guidelines for human annotators.
- Multilingual natural language processing: experience with processing English, French and Spanish.
- Information extraction: design, implement and deploy pipelines that process text and extract business-specific information (for example, to fill an Excel file).
- Marketplaces: matching products to costumers, jobs to applicants or friends to newcommers.
- Code migration: moving from unmaintained, legacy ML libraries to state-of-the-art technology.
- Machine learning (ML) in the browser: transforming ML solutions to run in the Web browser for lower cost and enhanced privacy.
- Machine learning operations (MLops) / data manipulation DSLs (domain-specific languages): designing and writing custom programming languages to handle specific data manipulation needs.
- Trainable distance functions: training functions that tell whether two items (customers, contracts, sales leads) are similar based on business needs.
- Automatic privacy cleaning: detecting PII (personally identifiable information) and replacing it with placeholders.
Other consulting firms are staffing firms. In our case, all work is done by Dr. Duboue and as such, availability might be limited at times. (If you need a team to develop a full solution, Dr. Duboue can refer you to such companies.) Currently, we have one slot available of 5 hs a week.
Dr. Duboue is located in Vancouver, British Columbia, Canada and works remotely at times compatible with Pacific Time. Once the current pandemic subsides, he is open to contracts that include scheduled trips within North America.
Dr. Duboue engages in work-for-hire contracts with complete IP transfer, billed by pre-negotiated hourly rates, payable by check or wire transfer to a Canadian bank. Canadian customers will get GST added to their invoices. His time can qualify for Canadian SRED credits as a consultant. During the contract, Dr. Duboue is available for work meetings at the negotiated rate. He has a limited number of consulting hours per week and limited number of concurrent contracts.
- Does not engage with competing customers concurrently.
- Has helped file patents on behalf of several customers.
- Other technology includes: Spark, Scikit-Learn, NumPy, OpenCV, Tensorflow, Keras, OpenAI Gym, Jupyter Notebooks, Node.js, Elm, Bootstrap, Flask, Scalatra, Django, Git, Bash, Linux (Debian and derivatives), Docker, Weka, Mallet, UIMA, spaCy, NLTK, Knowledge Graphs, SPARQL, RDF, JSON, XML, XSLT, Rule-based systems, RuTA, ClearTk, PHP, Perl, Latex, Lua, SQLite3, PostgreSQL, JNI, Delphi, VB, Win32. See his resume for more details.