Artificial Intelligence Engineering

As long as artificial intelligence solutions get popular, companies need to address the complexities of the design, deployment and operations of AI systems.

MLOps attempts to address some of the challenges, but presence of technical debt in data introducing complex dependencies, the model drifting or security risks required further analysis.

This short presentation does not intent to present specific solution, but surface the main challenges .

Transcript

1. AI Engineering Andrés-Leonardo Martínez-Ortiz, PhD Madrid, July 14th 2023
2. About me… Andrés-Leonardo Martínez-Ortiz a.k.a almo, holds a PhD on Software, Systems and Computing and a Master on Computer Science. Based on Zurich, almo is a member of the Google Machine Learning Site Reliability Engineering team, leading several programs aiming for reliability, efficiency & convergence. He is also a member of IEEE, ACM, Linux Foundation and Computer Society. @davilagrau almo Subscribe to my newsletter 👉
3. Let's start with some data* * Charlie Giattino, Edouard Mathieu, Veronika Samborska, Julia Broden and Max Roser (2022) - "Artificial Intelligence". Published online at OurWorldInData.org. Retrieved from: 'https://ourworldindata.org/artificial-intelligence' [Online Resource]
4. Complexity or how does AI engineering challenges loo like?
5. Increasing prioritization and budget ˜80% of the enterprises have AI as priority ˜83% of the enterprises increase the AI budget 49% process automation 45% managing logistics 43% supply chain optimization 57% customer experience 50% customer insights 48% customer interaction Increasing number of use cases Increasing saving and adoption speed thanks to specialized partnership IT partners allow savings up to 30% of the time IT partners demand ˜25% less of internal IT resources IT partners allow savings up to 20% of the budget Scale decreasing efficiency ˜40% of the enterprises expend +50% in development Increasing deployment time (+64% MoM) Maturity long curve 55% of the enterprises on evaluation or early stage of development and deployment Multiple & fragmentated success metrics make difficult the evaluation Software supply chain and configuration management diminish AI quality 36% of the enterprises have serious problem with model performance 56% of the enterprises have serious problem the security and auditing models 67% of the enterprises have to be complaint to several quality standards.
6. MLOps is a set of standardized processes and technology capabilities for building, deploying, and operationalizing ML systems rapidly and reliably.
7. Photo by Bia W. A. on Unsplash Technical debt Machine Learning cannot be effectively implemented in software logic without dependency in external data
8. Risk Modelling Photo by Edge2Edge Media on Unsplash
9. Risks Vulnerabilities in complex systems (hardwares, software, data) Deceived Human Interaction: misleading reporting Unpredictable sequential planning (collusion) Difficulties being shut down Photo by Tom Morel on Unsplash Deployment of ML systems required risk analysis, including technological, business and legal perspectives
10. Actively research is great, but… It is widening the engineering gap.
11. Maturity Frameworks Communications • Incident analysis and reporting, including external parties. • Auditability • Scientific peer-review • For business units and non technical staff. Security • Intensive evaluation strategies • AI-based monitoring • Fast responses protocols • Integrity verification, authorization and auditing Photo by Alexander Grey on Unsplash
12. Photo by Jan Tinneberg on Unsplash Thank you!