My AI Engineer degree covers not only data analysis, machine learning and deep learning, but also AI related cloud architectures and ethical considerations plus AI project management. To read more about the 11 AI projects developed along this path, and other former projects, please have a look at my 🏡 portfolio and resume (in french) . All these projects are also available on my OC AI Engineer Github (private) repository (please contact me and send me your github user name so that I can grant you a full access to these repositories)

🎯 In 2022 I started my fourth professional life when I graduated as 🎓AI Engineer (my field is also ML Engineer, Data Analyst, Data Scientist and Data Manager)! This is my natural evolution after my three former professional lives :

• (2005-2012) As a PhD, researcher and lecturer in 💻Computer Science and 💬Natural Language Processing.

  With research in Computer Science and NLP, I found my deepest calling and the activity in which I recognize myself the most. Find out more on Researchgate. 🤖Artificial intelligence is now fully a part of research and industry in NLP, enabling the design of solutions for everyday business and great future innovations. Beyond my former skills in symbolic linguistics, it was therefore necessary to learn how to design machine learning and deep learning solutions for NLP. It is a great asset now to master both approaches.
• (2006-2016) As a Fullstack developer, freelance then lead programmer in a few startups including my own.

  I have sold and delivered applications, web solutions and ERPs for various companies or organizations: in health, university, associations, small businesses. Thanks to these jobs, I have built a solid knowledge of organizational processes and business IT in the different departments of companies. These jobs allowed me to master the collaborative design and deployment of applications, from big databases to web or standalone front-ends, through complex server-side algorithms and electronic flows between systems. Data concerns, data structuration, data algorithms, were always present in my jobs. Personal data protection and ethical concerns in respect for patients and customers have always been in my mind constraints to deal with all along the software design process. Entering the era of data management, I am prepared to develop ethical and explainable solutions.
• (2010, 2018-2020) As a business-oriented project manager or information systems manager.

  I managed software bricks, electronic flows, IT teams and software developers for a few companies. I was often confronted with communication issues between departments and direction. Communication in the decision chain and user training are important bases to guarantee the quality of data, from its collection to its exploitation. The need to include data governance in the processes of organizations is thus a very clear imperative. For this reason again, my evolution towards a transversal job of data manager was necessary.

Taking the best from my previous experiences and recent training paths, I aim at becoming a better 🧙‍♂️ data wizard everyday. It takes long to be good in all aspects. I learn from each experience and a continuous watch ;)

• Identify business needs

  • More than 10 years working on software development for business and health
  • 8 years working with retail companies to improve their departments IT and information systems
  • I speak fluently with all business and industry departments to transform their needs into technical requirements
  • Each country has its own laws and rules, for commercial, supply or financial workflows. I already dealt with such differences in France, Germany, Italy, Algeria, Morroco, UAE and China. Eager to learn more how things are done in more countries!

• Set up data architecture

  • As software project manager and as Information Systems manager, I am used to design and deliver data flows, databases, services, web applications and ERPs.
  • I'm familiar with on premise (linux or Microsoft servers) and Azure cloud solutions.
  • Some of my recent projects were directly devoted to deploy Azure ML applications (serverless, functions, cognitive services, pipelines with ML studio, online training and continuous deployment via github actions)
  • I'm also comfortable with various data sources and electronic flows: on premise or cloud databases, API, flat CSV or structured EDI (EDIFACT for instance), ETL, versioning, normed flows. I'm using batch processing to handle complexity issues with big datasets.
  • Business-specific data formats and norms are important not only for internal reliability purposes, but also to communicate with external services and systems.

• Analyse data

  • Ability to analyse, model and interpret data
  • Accuracy and attention to detail
  • Harvest, clean and explore data
  • Mathematical sharpness, statistics and probabilities, metrics
  • Data wrangling
  • Feature engineering
  • Methodical and logical problem solving approaches
  • Univariate and Multivariate analysis
  • Dimension reduction
  • Data visualization and visual exploration are fundamental for end-users in all business departments: from exploratory insights to key performance indicators, via visual dashboards for any team.

• Write production-ready code

  • I have often dealt with critical production delivery concerns, with realtime or uninterrupted services constraints, for worldwide customers or health-care patients handling.
  • In a collaborative workflow, developers, testers and beta testers are working on several versions of the same applications, each having its own environment.
  • Software and Data operators have to work hand in hand with IT, system engineers and product owners to handle maintenance and production delivery schedule
  • Managing a software or data project requires developers and analysts to document their code and data.

• Build ML models

  • I use libraries such as Scikit-learn and Keras, with Pytorch or Tensorflow.
  • Transfer learning with HuggingFace or other open models
  • I use train-test-split and other folding techniques to tune hyperparameters
  • Bias avoidance strategies via dataset balancing
  • I expose and explain models outputs with techniques such as features importance calculation
  • I calculate and define model periodic maintenance schedule, according to model deprecation metrics based on the business cases
  • I build models locally or in cloud (MS Azure)

• Deploy models into production environment

  • I deploy models in local area networks, servers or in cloud (MS Azure)

• Automate processes

  • Production delivery with orchestration tools or with cron tables and scripts.
  • Collaborative continuous integration and delivery (CI/CD) with Github on-push Actions

• Present results

  • Documentation
  • Progression reporting to product owners and stakeholders, with agile / scrum / stories best-practices in mind
  • I design dashboards according to business departments requirements
  • I develop data visualization and visual exploration frontends for several users according to their jobs and requirements: Jupyter Notebooks, Voilà!, Streamlit, Flask
  • Further on, I am also keen on publishing R&D results in academic or business contexts

• Governance and transversal Project Management

  • A project kicks off and goes live with several services, hierarchies and stakeholders working together. I use Agile management, with Jira and Confluence, scrum, sprints, to align transveral requirements with time to market concerns
  • Budget, security and ethical aspects need to be handled by decisional actors working together, as well as GDPR aspects. Data management has to be transveral even in small companies, rather than hierarchically enclosed in an IT subdepartment

• 👀 I’m interested in open source collaborative development since my first professional software development in 1997. What a time to be a developer! When I wrote my first basic program in 1984, open source code was available in specialized newspapers. Now no-code and continuous delivery are realities. So what is still accurate after 37 years in programming computers? As far as I was involved in critical algorithms, 🤯the ability to optimize memory and calculation complexities remains a HUGE skill.

• 🤝 I have some open source collaborative projects in progress on the following topics :

  • 🔭 Courses and e-learning exploration to gain bird-eye-view over training paths for data science in particular and any kind of diploma in general. See OCCoursesExplorer and a demo
  • 🦄 Skills management and ontologies to help matching resumes and job offers.

• 🌱 I’m currently discovering very interesting ML models with 🤗 HuggingFace!

• 💬💖 NLP R&D remains my beloved field

  • I care for symbolic Natural Language Processing: morphosyntax, syntax, dependencies and syntagmatic grammars, constraints formalisms, metagrammars, lexicology, corpus linguistics, semantics, ontologies.
  • With AI and Machine Learning algorithms, NLP can move to an industrial level, with AI language models, topic modeling, sentiment analysis, image classification, text-to-speech and speech-to-text processing, and plenty other business cases such as chatbots, translation, information retrieval in textual resources. Some of my recent projects (in french) were dealing with these cases.
  • 🎯 In the first age of AI, some NLP expert-systems based on deductive inference techniques were claimed to be artificially intelligent. However they were what one calls mirrors automata mimicing human language with low robustness and poor efficiency in fuzzy situations. These systems had at least an ambition to approach human reasoning skills. In todays AI, NLP actors are claiming to achieve intelligent tasks, with higher robustness and tolerance to lacunar information, but these tools remain compared to speaking parrots unable to process any logical deduction or to account for contextual realism. 🏹 I would like to get involved in hybrid research projects where deduction, inference, and calculation, work hand in hand with AI language models, moving AI systems closer to intelligent reasoning.

• 🤟 Topics I enjoy as a hobby:

  • 🎸🎹 composing and playing music
  • 🖌 drawing and painting
  • 💎 stone balancing
  • ☕ reading and writing technology watch content

• ✍ Recent data science posts on medium.com

• 🏡 My portfolio and resume (in french): https://tristanv.github.io
• 🤝 My collaborative space for public and private projects : https://github.com/TristanV
• 🔗 My professional network : https://www.linkedin.com/in/tristan-vanrullen/
• 📜 My academic publications and research home page: https://www.researchgate.net/profile/Tristan-Vanrullen
• ✍ Some data science publications : https://tristanv.medium.com/

This page was updated on 2022-12-18.

Visits since 2022-12-18 :