Mendel.ai creates novel artificial intelligence systems to assist cancer research. The AI powers software for hospitals and medical research organizations to accelerate and improve clinical trials. Our AI has been recognized as the most sophisticated of its kind.
We're on the cutting-edge of developing software that will directly impact cancer research and patient care.
Our AI works. In a controlled experiment conducted at a cancer clinic, it took the staff hundreds of days to identify 50% fewer patients than our AI did in minutes. The patients were being identified for eligibility for clinical trials. All the results from the experiment were published in a peer-reviewed paper.
We prize creativity and interesting ideas, and to foster this, we create a dynamic, anti-bureaucratic work environment.
We are growing quickly, backed by visionary investors who fully support our mission and unorthodox approach.
We have two primary engineering teams, one team that works on web applications in React and Node.js and one that works on clinically driven AI projects in Java, Python, and C++.
Engineers in the web development team work according to the Agile methodology and coordinate with both the DevOps and UI/UX team to meet all product requirements.
Engineers in the AI team work according to the Waterfall methodology by fully researching and implementing solutions according to pre-defined specifications.
OCR (Optical Character Recognition) Challenges:
How can an AI transcribe text in an image to a machine-readable form?
Medical document images are notoriously noisy. After being faxed time after time, these documents are riddled with excess ink, and other noise, that makes transcription very difficult. Further complicating matters are the many rare, very long words that fill these documents, e.g. 'hepatosplenomegaly.' The longer and rarer a word is, the more likely the machine is to read it incorrectly.
NLP (Natural Language Processing) Challenges:
How can an AI parse medical language in the way a human would?
It needs more than stock NLP engines; a specialized conceptual framework is required to parse the text according to medical classifications like diagnosis, intervention, biomarker, etc. When the end-goal of parsing medical language is evidence, there are certain ambiguities, which if misinterpreted, can ruin an analysis. For example, how can the machine determine whether the phrase
is consistent with signifies a possibility or an actuality when it is used in both cases?
Knowledge Representation Challenges:
How can an AI acquire knowledge of a patient that's explainable to a human?
Machine learning is a black box, but this won't fly in medicine. Any facts or analyses the AI offers as evidence must be perspicuous to a human, i.e. the human must be able to understand the reasons behind the AI's output.
Be Realistic: Extravagant claims can kill patients. At Mendel, we tell the real story about the potential for our AI and its impact on healthcare. We are committed to the truth and honest assessment.
Lead With Originality: What we are creating is an original solution. We strive to innovate with every iteration and to do things on our own terms, not copy others.
Commit to Excellence: We never compromise quality, and we are committed to cutting through confusion. Real-world, quality data are the core of our business.
Make a Difference: Our decisions are driven by data. We don’t “wing it.” We are committed to getting it right and use data to support our decision making, leading to more insights that could potentially change the world in healthcare.
Think Like a Physician, Code like a Developer: We are physicians first, technologists second. All of our solutions are in the service of patients. With a custom-built AI system, we are committed to combining an oncologist's knowledge and intuition with a machine's precision.
Greet Challenges Head On: We don’t shy away from challenges. We began our journey tackling the most difficult and complicated medical field: oncology. Today, we continue to take on complex medical challenges with vigor and enthusiasm.
Interested in this company?
Skip straight to final-round interviews by applying through Triplebyte.