Senior tech executive with a decade experience leading large, globally
distributed engineering teams, and turning innovation into material
business impact.Strong technical background in computer vision, mapping,
geospatial imagery, machine learning, large scale systems, image
processing & analysis, document imaging, and more. Equally effective
in startup or large company environments.Experienced leading cross-functional efforts involving software
engineering, hardware development, product management, partnerships,
operations, finance, and more.
Vice President, Engineering
Lead Marketplace & Autonomous Platform division (MAP), which includes
in particular Lyft's ``Level 5'' group, focused on Self-Driving
Technology. The Level 5 team is building an open autonomous
stack--software and hardware--to complement the Lyft Open
Platform. Target is to launch self driving vehicles on the Lyft
network, alongside with AVs from our network partners.
The MAP team is also working on a range of super interesting technologies
such as accurate real-time location, routing, supply/demand matching,
predicted and real-time traffic, event prediction and detection,
multimodal navigation, crowdsourcing, mapping for human-driven and
autonomous vehicle, and more.
My career at Google has focused on products and platforms based on the
collection and organization of geographic imagery at global scale. The
most well-known is Google Street View, but it is only one of the
many projects and products I have been responsible for starting and
scaling over my 12 year tenure with the company.
Senior Director, Engineering (2012-2017)
Led globally distributed Geo Imagery organization including several
hundred engineers, product managers, operations managers, technical
program managers, and other functions, plus a much larger number of
temps and vendors. Org is headquartered in Mountain View, CA, but
includes (or has included) teams in Boulder, Seattle, Los Angeles, New
York, Tokyo, Beijing, Paris, Zürich and Munich. Mission is to (1)
efficiently capture petabytes of imagery around in the world, from
cars, boats, planes, satellites, web crawls, power users, etc; (2)
organize pixels: georeference, align, stitch, blend, orthorectify,
3D-reconstruct, recognize objects, index geographically & by content,
and more; (3) use imagery and derived semantics to power a range of
Google platforms, data quality efforts, end-user products (eg, Google
Maps & Google Earth) and emerging initiatives.
- Scaled Street View to over 80 countries. In addition to being
one of Google's best recognized brand assets, Street View
remains a uniquely differentiating feature of Google Maps.
- Led Google's efforts to leverage deep learning in order to
create high quality map and local data globally, from imagery and
- Bootstrapped and scaled efforts to capture high-resolution
aerial oblique imagery and create 3D models of the world's top
metropolitan areas. Our efforts have far outpaced the competition in
both geographic coverage and quality.
- Successfully integrated Skybox Imaging ($500M
acquisition, focused on small imaging satellites) in 2014; now
rebranded as Terra
Bella. Team built outstanding tech
stack, successfully launched 5 sub-meter resolution
satellites and was on track to launch many more, before
Google decided to get out of the
- Led multiple camera-centric hardware efforts, from initial R&D
through production and deployment at scale.
- Responsible for several apps and consumer products, including
Google Earth, Street View App and more, across Web, Android
and iOS platforms.
- Led efforts to define, build and scale a Geo platform business,
in collaboration with Google Cloud team.
Director, Engineering (2008-2012)
Led Mountain View, CA, and Zürich-based engineering team responsible
for scaling Google Street View. Drove efforts to apply the
Street View ``formula'' to adjacent areas (eg, aerial/satellite
imagery), thereby reducing engineering complexity and
fragmentation. Championed and grew Google's efforts to derive
exponentially more value from these global imagery assets.
- Grew Street View from a US-centric experiment into a
critically important Google product, globally. Street View was
recognized as the fastest growing product at Google in 2008, and has
continued to grow strongly ever since.
- Took Google's aerial imagery acquisition efforts from their
``garage phase'' to meaningful scale by applying the same
disciplined engineering approach used for Street View; merged
projects/teams related to camera hardware, onboard software, field
support, leading to significantly improved velocity, efficiency and
- Continued to drive innovation in a number of areas: eg,
introduced VR-like ``Smart Navigation'' in Street View by leveraging
extracted 3D data; bootstrapped efforts to recognize objects in
imagery at scale, to refine understanding of the physical world.
- Engineering sponsor for a string of Google acquisitions. In
particular, I co-led the acquisition of reCAPTCHA and I have
played a key role in growing it into a highly successful Google
product (to date I am still responsible for reCAPTCHA).
Uber Tech Lead / Engineering Manager (2004-2008)
Responsible for several engineering aspects of Google Book
Search. Head of Google OCR-related initiatives. Founder and leader
of multiple novel engineering efforts, including Street View.
- Led efforts to grow Google Book Search indexing to dozens
of new languages.
- Instrumental in ``resurrecting'' the Tesseract OCR
package, open sourcing it, and leading its eventual transformation
into one of the world's best OCR engines.
- Started Street View project in ``20% time''; hired an
all-star team of software engineers, computer vision scientists and
operations specialists; successfully launched product in May 2007,
and turned it into an iconic Google success story.
LizardTech, Inc. (LTI) was a software company focusing on the
compression and efficient distribution of complex imaging content.In 2000, the company acquired the DjVu document compression and
delivery technology from AT&T Labs, which is why I decided to join
LTI. Despite our efforts, DjVu never gained broad acceptance,
but a decade later it remains state-of-the-art for scanned document
compression and representation.
Vice President, Document Imaging (2003-2004)
Responsible for all aspects of the LTI document imaging product line,
including, marketing, strategy, product management, partnerships, etc.
- Overhauled document imaging strategy following Company's
acquisition by Celartem Technology.
- Led aggressive product development efforts to support new
strategy with key features and product improvements. Drove design
and implementation of new watermarking and document rights
management (DRM) capabilities.
- Focused LTI's marketing efforts on high-leverage activities,
including a rebooted web site, regular newsletters and webinars,
targeted interviews and selected trade shows.
- Negotiated partnerships with scanner vendors, content management
providers and other technology partners, in a bid to enhance the
company's distribution channels and to increase end-user exposure to
the DjVu technology.
Chief Scientist, Document Imaging (2001-2003)
Share time evenly between technical activities, product development
and business development, for the DjVu product line.
- Chief company-wide technical resource on document imaging.
- Led efforts to expand product line to electronic document
conversion via virtual printer driver and other tools.
- Embedded with product management team to support all aspects of
document imaging product line: feature sets, UX, platforms,
competitive analysis, schedule, pricing, licensing, documentation,
packaging, distribution, etc.
- Relentlessly evangelized DjVu technology via conference
presentations, publications, and the creation of compelling DjVu content.
- Identified, negotiated and closed deals with several strategic
Director of Applied Research, Document Imaging (2000-2001)
Led DjVu Applied Research team while remaining active individual
- Responsible for large portions of company's core imaging
- Championed and led new R&D initiatives focused on growing scope
of product line and size of addressable market.
- Key contributor to company strategy, and proactive business
- Advocate for LTI's continued involvement in open-source
The Adanced Systems Development Lab (ASD)'s mission was twofold: help
Xerox realize the value of PARC research through effective technology
transfer; bring business group needs and perspective back to PARC.
Xerox Palo Alto Research Center (PARC), Advanced
Systems Development Laboratory
Manager, Imaging Components & Services
Led a team of 8 researchers and developers. Supported ASD's mission in
the area of imaging through development of rock-solid sofware
components. Collaborated with other PARC teams and coordinated efforts
with geographically distributed Xerox business groups.
- Led development and deployment of DataGlyph technology
(high density, high reliability, visually pleasing 2D barcodes) in a
range of Xerox solutions.
- Managed development and productization of a technology known
internally as DigiPaper. Core of technology is a ``visually
lossless'' scanned document representation based on multi-layer
segmentation, coupled with wavelet-based and token-based compression
schemes. Released DigiPaper 1.0 in May 2000, for use in web-based
legal document repositories.
ScanSoft, a document management software company, was born out of the
merger of Xerox Imaging Systems (XIS, a subsidiary) and a Xerox group
called Software Solutions Division (SSD). In 1999, Scansoft went
public through a ``reverse merger'' with Visioneer and in 2005 it
merged with Nuance and took its name.
Director, Advanced Development, ScanSoft (Palo Alto, CA, 1997-1999)
- Led focused engineering team responsible for advanced image
processing development, in support for ScanSoft's entire product
line. In particular, team developed core of Pagis Pro, a
scanning software package that received a number of awards and
became the best selling scanning suite on the market.
- Orchestrated development/productization of novel technologies
for scanned document management: binarization, adaptive compression,
automatic cropping, noise cleaning, deskewing, enhancement,
superresolution, etc. This played a key role in differentiating
Scansoft's product offerings as well as building up the company's IP
- Promoted and put in place new processes for improving
engineering productivity and code quality: documented requirements,
engineering reviews, version control, code reviews, automatic
builds, system tests, documentation, etc.
- Kept close ties with Xerox PARC and other research organizations
to ensure steady flow of new ideas to ScanSoft.
Director, Software Development, Xerox Software Solutions
Division (Palo Alto, CA, 1996-1997)
- Led Image Processing Core Technology Group (IPCore), a team
delivering cutting-edge document image analysis software to a
diverse base of internal customers (software and hardware product
teams). Oversaw release of about a million lines of C code,
on five different platforms (Windows NT, Windows 3.1, Macintosh,
- Responsible for handling SSD's relationship with Xerox research
groups, including PARC and the European Research Center (XRCE).
Managed $4M budget for contracting specific technology pieces with
these research groups. Facilitated tech transfer from Xerox research
into SSD, and coordinated efforts to steer PARC and XRCE into
specific research directions.
Manager, Image Processing and Analysis, Xerox Imaging
Systems (Peabody, MA, 1995-1996)
- Managed team of 2-3 engineers focused on developing document
image analysis algorithms for the TextBridge OCR package.
- Bootstrapped XIS government contracting business and was awarded
$0.5M contract by National Security Agency to develop new
methods for benchmarking page segmentation algorithms. Served as
Principal Investigator for this project over its two-year span.
Member of Technical Staff, Xerox Imaging Systems (Peabody,
- Singlehandledly developed page segmentation library that powered
XIS's entire product line, from shrink-wrap software packages (e.g.,
TextBridge) to reading machines for the blind (The
Reading Edge). System worked reliably even for complex layouts
and was rated most accurate segmentation package on the market at
the 1995 UNLV OCR accuracy contest (UNLV studies were a reference in
the Document Recognition world).
- Major contributor to several other pieces of XIS' document
recognition technology, including character recognition, page
filtering, binarization techniques, algorithmic and architectural
Harvard University, Robotics Laboratory
- Published and generalized morphological algorithmic techniques
developed during my PhD work.
- Initiated collaborations with some of the best researchers in
the field of computer vision.
- Participated in design of DIP Station, an image analysis
package for the Mac.
- Developed novel technique for assessing corneal health from
corneal endothelial tissue images (in collaboration with Uniformed
Services University of the Health Sciences).
Associate Research Fellow
École Nationale Supérieure des Mines de Paris,
- Worked towards my PhD degree, under the guidance of Prof. Jean Serra.
- Helped promote image processing expertise of the Center for
Mathematical Morphology, and land research contracts with Aérospatiale and the French Institute for Petrol (IFP), for
which I acted as lead investigator.
In the 1990s and early 2000s, I had a lot of very interesting
consulting assignments on the side, in fields including medical
imaging, remote sensing, industrial inspection, optometry, and oil
exploration. Some key customers and projects include:
- Atlantic Aerospace Electronics Corporation, Waltham MA
(1991-93). Developed new algorithms for automatically extracting
correlogram tracks in sonar imagery. Algorithms are capable of
detecting tracks under extremely low signal-to-noise ratios, where
previous approaches failed. Also designed hierarchical
watershed-based algorithms to robustly extract and recognize targets
in Forward-Looking Infrared (FLIR) imagery.
- Woods Hole Oceanographic Institution. Woods Hole, MA
(1992-99). Participated in the design of a system to automate the
detection and classification of plankton from towed video microscopy
sequences. Provided some of the key segmentation and feature
extraction algorithms used in the system.
- Vysis, Inc., Downers Grove, IL (1992-96). Developed
advanced segmentation and pattern recognition algorithms aimed at
chromosome analysis from blood sample.
- Chevron Petroleum Technology Company, La Habra, CA
(1996-2004). Proprietary project involving oil exploration and
requiring the processing of huge 3D data sets.
- Humphrey Systems (now Carl Zeiss Meditec), Dublin, CA
(1998-99). Developed key image enhancement and feature extraction
algorithms for ophthalmologic instruments.
- Applied Imaging, Santa Clara, CA (1999). Designed novel
methods for discriminating between different classes of cells in
Ph.D. in Morphological Image Processing,
École Nationale Supérieure des Mines de Paris, France, 1990
Masters (``Diplôme d'Études Avancées'') in Computer
Science, Paris XI University, France, 1987
B.S. (``Diplôme d'Ingénieur''), Majoring in Mathematics and
Computer Science, Ecole Polytechnique, France, 1986
- Author of 60+ publications on computer vision, image
processing, mathematical morphology, OCR, document understanding,
and related topics. See www.vincent-net.com/luc/papers/.
- Keynote speaker at dozens of industry & academic events,
including ISVC'15, DASFAA'15, IVRPA'15, SIGSPATIAL'14, ICCV'13,
WACV'11, ION GNSS'11, ICDAR'07, and many more. Also invited to speak
at companies & institutions around the world, on topics ranging
from vision research to product development to leadership.
- Supporting the Research Community:
- Research grant reviewer & sponsor: primarily while at
Google, I have reviewed dozens of grant applications and been
directly or indirectly responsible for millions of dollars in
- Associate editor of the IEEE Transactions on Pattern
Analysis & Machine Intelligence (PAMI), 2000-2004, and of the
Journal of Electronic Imaging, 1994-2002.
- General chair or co-chair for dozens of conferences since
1991, including: SPIE conferences on Document Recognition, Visual
Communications & Image Processing, Nonlinear Image Processing; IEEE
conferences on Computer Vision & Pattern Recognition; IAPR
Workshops on Document Analysis Systems; International Symposiums on
Mathematical Morphology; ICDAR conferences; etc.
- Peer reviewer for several journals since 1988,
including: IEEE Trans. on Pattern Analysis & Machine
Intelligence, J. of Visual Communication and Image
Representation, Signal Processing, IEEE Trans. on Acoustics,
Speech & Signal Processing, IEEE Trans. on Systems, Man &
Cybernetics, IEEE Trans. on Medical Imaging, IEEE Trans. on
Image Processing, Acta Stereologica, Journal of Mathematical
Imaging, Journal of Electronic Imaging, etc.
- Patents & IP:
- 71 issued U.S. patents (as of 03/2016), plus 29 applications
pending. See www.vincent-net.com/luc/patents/.
- Over a 5 year period at Google, led my product area's efforts to
significantly increase patent filing rates. We were consistently
ranked as one of the most prolific group in the company.
- Deposed on a few occasions in relation to patent litigation cases.
- Citizenships: USA, France
Lisp, bash, csh, Matlab, Apache, LATEX, etc. Experienced Linux and
- Languages: English (fluent), French (native), German (was
excellent; now rusty)
- Hobbies: home automation, VR photography, adventure travel, rock
climbing, backpacking, road biking, backcountry skiing,
mountaineering (successfully organized and led expedition to summit
of Denali, Alaska in 1991), running the IT/scoring infrastructure at
mens gymnastics competitions.