Eng. Cristian Vicas, PhD
Navigating the AI hype since 2008
More and more companies employ Machine Learning to solve really hard problems.
Searching for the right solution usually consumes a lot of time and engineering efforts because the field is full of noise and hype.
As a freelancer I use the lean approach to find a suitable solution and adapt it to the problem in hand.
All these, while keeping the ML exploration time and budget under control.
Move fast from idea to production
Your customers’ desires and the market in general is always on the move. To be relevant, a company must be nimble. I follow the lean approach so I can iterate quickly on various technical approaches. Each iteration will end with a test “in the market” so the ML development is directly driven by market feedback.
Keep the costs under control
Creating the perfect solution is a luxury that few can afford. It takes a lot of time and resources and there is always the risk of building the wrong thing. By focusing on one issue at a time, and always taking the customer’s feedback into account, we can move towards that perfect solution while constantly delivering value to the customer!
Machine learning is not the only answer
Most of the time Machine Learning is only part of the solution. Other aspects of the problem can be easily solved with solutions from hardware or web/mobile applications. I have technical skills in these areas to know their limitations and to recognize when such a solution must be added. Sometimes Machine Learning is the wrong answer altogether.
Just want to try it out?
You need to validate some business hypotheses? Want to test a risky approach? You will want a solution delivered cheap and fast! My fast prototyping abilities and out-of-the-box thinking will deliver solutions that will allow you to circumvent the roadblock and allow the project to move forward in a timely manner. Sometimes a decent solution delivered now is far better than the perfect solution delivered in a few years.
Your problem seems unsolvable and requires a deep dive and hard focus?
You need a partner that is resilient and not afraid to fail. Experimenting is key in machine learning and the tougher the problem, the more failed experiments will be needed. I can explore solutions for a long period of time until your resources run out, at each step providing a thoroughly documented list of failed solutions, their shortcomings and promising future directions. The priority will always be to improve, at each step.
You will want somebody with strong communication skills that can extract and communicate the essence of a problem. One key ingredient that I used when working with experts from healthcare, space, agriculture or heavy industry was my ability to learn their “language”. In this way I am able to understand their pain points and explain back potential solutions
Navigating the scientific literature requires a skeptic mind that can “read between the lines”. During my academic years I’ve read, written and scientifically reviewed many many papers. I still perform occasional scientific reviews so I am up to date with cutting edge research, before it reaches the public audience.
Sounds familiar? Let’s have a chat!
Projects
LLM for Education
Industry and Research
Mentoring is important for learning but the mentor’s attention is divided between students. I developed a LLM powered tool that aims to help first year students gain clean-code experience.
Applied Computer Vision
Industry and Research
The bulk of my activity is focused on computer vision. These are some questions that I am addressing. Of course, different data, different contexts.
Earth Observation
Industry
Earth observation satellites generate a lot of data and most analytic solutions are automated. I briefly touched two sensitive areas of applicative research in this field, the cloud detection and land labeling.
Medical Image Processing
Research
During my PhD my research topics were focused on medical imaging, with two main modalities: ultrasound and visible spectra. Contributed to six grants and over 40 papers.