I am an Associate Professor of Computer Science and Machine Learning at HES-SO Valais.
I am also a former Distinguished Engineer at
Cisco, where I pioneered AI/ML at Cisco with 4 flagship products at the intersection of AI
and large-scale software and distributed systems, such as cybersecurity, enterprise networks, and the Internet. These products
have now more than 5000+ enterprise customers, including many Fortune 50/500.
My focus is on fostering scientific, engineering, and operational excellence across my organization, especially in the area of
system design, data science, machine learning, and software engineering. I work best at the intersection of architecture, business,
engineering, and technology, where I can influence and align vision, architecture, roadmap, research, and execution.
At Cisco, my mission is to deliver disruptive (non-incremental) innovations, leading through their entire
lifecycle, from ideation, hiring, R&D, product design, architecture, and intellectual property protection,
to engineering execution and first customer shipment.
I co-architected and lead the development of 4 key products at Cisco:
Predictive Networks (2022) is the core technology underpinning several products (e.g.,
ThousandEyes WAN Insights
,
Cisco SD-WAN Analytics
) that brought for the first time learning capabilities to enterprise networks and the Internet. Routers
proactively redirect application traffic away from paths that have a high probability of experiencing
disruptions, thereby improving significantly the application experience of the end user. See
Predictive Networks at Cisco
for further details about the technology.
Cisco AI Network Analytics (2019) is a cloud-based machine learning platform that provides
visibility and learns from 2500+ enterprise networks across the world. Upon its launch, the product was
the first one to perform adaptive, ML-driven network assurance. See
this blog post
for further details.
Cisco AI Endpoint Analytics (2020) is built on the same platform as Cisco AI Network Analytics, the
product protects organizations from ransomware with ML-based anti-spoofing technology. Using behavioral
modeling and trust analytics, we detect devices that exhibit unusual patterns of communication typical of
a rogue device attacking the network. See
this blog post
for further details.
Self Learning Networks (2016) is a cybersecurity product capable of zero-day attack detection by
turning every router and switch into a security agent that ingests locally NetFlow telemetry, model the
behavior of every device in the network and trigger alerts in case of anomalies. SLN uses a
hyper-distributed learning architecture capable of processing up to 100 billion events a day across 5000
routers. Now part of
Cisco Secure Network Analytics
.
Team Leadership
I built 3 engineering teams built from the ground up. This amounts to more than 50 engineers hired, coached,
and managed across a broad range of geographies and disciplines (software engineering, machine learning,
data science, cloud engineering, data visualization, front-end design).
Software Development
I have a broad expertise in managing agile teams to deliver reliable software using continuous integration
and delivery strategies, test automation, code reviews, pair programming, and cloud devops. I wrote a
popular handbook that shows why I believe that building
software that meets both customer requirements and deadlines is possible.
Customer Impact
I lead the design and development of five products at Cisco used by thousands of enterprise customers,
including many Fortune 50/500, from inception to customer delivery in production. I attended dozens of key
customer meetings over the years, both to pitch our vision and technological capabilities, and to understand
their pain points and adapt our solutions accordingly.
Technological Breadth
My 10-year tenure at Cisco allowed me to gain insights into a broad technological landscape, including
cybersecurity, smart grids, Internet of Things, wireless networks, internet architecture, endpoint
classification, and cloud computing. Prior to that, I had worked closely with micro-engineers on the design
of Micro-Electro-Mechanical Systems (MEMS) and with chemists on the self-assembly of large molecules. I also
have a general interest in the area of climate science, energy, and sustainability.
Key Skills
Every now and then, I read again Matt Might's
'What every computer science major should know'
and I realize that I will never quite live up to its standards. Yet, I found it to be an excellent North
Star for my lifelong learning plan. The list below is a compilation of topics that I have seriously used
and/or studied at some point in my career, but some might not be so fresh, especially the theoretical
part (they are marked with an asterisk).
Programming languages:
Python, Polars, JAX, SQL, Apache Spark, C++
Software engineering:
data-intensive software, distributed computing, large-scale software systems
Technologies:
LLMs, machine learning, statistics, networking, robotics, data science
Experience
Associate Professor
HES-SO Valais
I am working on bridging the gap between state-of-the-art AI/ML methods and real-world applications,
working with many companies and research partners in Switzerland and beyond. I also teach classes in
computer networks, machine learning and data engineering.
August 2024 - Present
Distinguished Engineer (Senior Director)
Cisco
Delivering best-in-class machine reasoning and code generation for detection and remediation of IT
infrastructure issues powered by Large Language Models (LLMs), reinforcement learning and differential
programming (JAX).
As a Distinguished Engineer, I strive to live up to
these standards.
October 2022 - July 2024
Principal Engineer (Director)
Cisco, Predictive Networks
Architect and global engineering lead of the internal startup of 25 engineers that developed Predictive
Networks. This technology now underpins two products (ThousandEyes WAN Insights and Cisco SD-WAN
Analytics) and leverages Cisco's unparalleled visibility into the Internet to endow enterprise networks
with predictive capabilities. We worked with amazing customers across the world, including many Fortune
50/500.
January 2019 - September 2022
Head of Product Development (Senior Leader)
Cisco, Enterprise Networking Group
I was the Senior Technical Leader in charge of the machine learning group, as well as the development of
the
Cisco AI platform
, a cloud-based machine learning platform that provides visibility and learns from enterprise networks
across the world. I led the R&D process from white-boarding to the completion of first customer
shipment. Today, it is routinely used by 3000+ customers across the world.
My role was a highly technical mix of research, system architecture, algorithm design, software
development, product, and engineering management. I have hired and managed a multidisciplinary team of
12 software engineers, machine learning specialists and data scientists.
July 2016 - December 2018
Head of Machine Learning (Leader)
Cisco, Security Business Group
I was the Technical Leader in charge of the machine learning and software development group for
Stealthwatch Learning Network (SLN), Cisco's next-generation cyber security platform, now known as
Cisco Secure Network Analytics
. This product turns every router and switch into a security agent capable of analyzing the behavior of
every host and device on the network and triggering alerts in case of anomalies.
May 2014 – June 2016
Senior Software Engineer
Cisco, Internet of Things Group
Design and development of many machine learning algorithms, visualizations, and data analytics pipelines
for Low-power Lossy Networks (LLNs).
Technical project lead, architect and main developer of the first service assurance prototype for the
Internet of Things.
Lead developer of the very first prototype of Edge Computing at Cisco.
Hired and managed a small team of 3 software engineers.
My teams have always thought that building software that meets both customer requirements (by solving a
key problem) and deadlines is possible. We have been historically pretty successful in doing so, and we
strive to improve even further. This handbook introduces the foundational ideas and principles that have
helped us in the past, and that we are committed to live by in the future.
There is a subtle, yet fundamental reason why automation is a key ingredient in the success of AI in
networking and many other areas: causality. Note: the article is divided into
two parts .
I wrote this guide based on my experience managing various remote teams at Cisco. It provides practical
pieces of advice for engineering teams on how to communicate effectively, in particular on technical
matters.