– Big Data/AI & Cloud Architect (consultant)
May 2015 – Present
• Solution/Functional Architect and design lead for cloud, HPC and big data/AI initiatives at Danone
Research (Clinical, Specialised Nutrition, Waters, Marketing/Analytics).
• Identification of new opportunities, selection and review of industry solutions, coordination with
software vendors (Cloudera, AWS, Microsoft, Tibco, Alteryx, Databricks, Rstudio, Siemens).
• Projects:
- Multi-regional Datalake using Cloudera and AWS services,
- HPC setup for computational fluid dynamics and computational chemistry workflows,
- DataScience and deep learning for nutritional impact, medical image recognition, biomarker
diagnostics by medical smart devices, specialised nutrition statistics.
- Video Intelligence platform using AWS services and LLM/Generative AI libraries.
• Code review, training/enablement and best practices for data scientists/developers and IT support.
• Certifications: AWS Architect-Associate, GCP Architect-Professional, DATABRICKS LLM
Professional
AWS, GCP, AZURE, CLOUDERA, DATABRICKS, ALTERYX, RSTUDIO, HPC, STARCCM/+CULGI, , LARGE LANGUAGE
MODELS/GENAI MODELS
Consultant, Self-training
juin 2014 - avril 2015
Speech synthesis project; Research and development of a voice adaptation module, improvement of natural language processing (NLP), studies for integration on iOS.
Preparation of a general architecture document and roadmap for the usage of Big data in urban transport systems.
Certification: Machine Learning (Stanford).
Sophis post-trade downstream simplification project: improvement on scalability, multi-versioning, unified and optimal data access mechanism on Sophis systems using OLAP or in-memory database. Definition of the XML data model for the description of financial products. Architecture review and Gap analysis. Organisation of workshops and steering committees with different stakeholders. Delivery of a general architecture document and a project plan.
CWT: Workshops with data scientists to identify the requirements, delivered an architecture document for a small cluster, sharing Hadoop and Cern Root. Training on map/reduce and best practices to the dev team.
Grid Architect
IPM – (Institute for Research in Fundamental Sciences, Tehran, Iran - ********)
avril 2013 - octobre 2013
National Grid Initiative, benchmark for Glite grid clusters, POC on distributing Matlab instances on PrActive grid.
Environment: Glite, High performance Linpack, Quantlib, C++, Linux, ProActive, Matlab.
– Grid Architect (consultant)
HSBC – London, Paris
avril 2012 - mars 2013
Review of Valuation Service: Fixed-Income workflow server for distributing pricing calculations on a Datasynapse grid and using Oracle Coherence for data management. Improvements on data caching and distribution of calculations, design of a data fabric, detailed benchmarking on client, middleware and quants code. Specific use case study for the CVA/PFE batch.
Gap analysis, capacity planning, design of monitoring reports.
Delivered detailed comparison reviews between Datasynapse Gridserver / Platform Symphony.
Performed review and analysis of ActivePivot in memory OLAP system as a general solution for risk platforms within Fixed Income.
IPM – (Institute for Research in Fundamental Sciences, Tehran, Iran - ********)
novembre 2011 - mars 2012
Worked on the setup and design of a National Grid Initiative: Federation of data-centres from different universities using CERN Glite midlleware for a mutualised grid infrastructure.
Gave lecture at an important HPC seminar: presentation on "grid computing problematic in financial risk calculations. Map/Reduce, Hadoop, HDFS detailed architecture” (slides: ********).
Environment: Glite, Apache Hadoop, HDFS, Linux.
Software Developer & Architect, Risk System (permanent)
BNP-Paribas UK
décembre 2010 - octobre 2011
DynaBatch – Design/Development on the new Risk System unifying counterparty and market risk projects, using Compute and Data Grid solutions.
Advisory, optimisations and reorganisations on C++ code.
Studies on Map/Reduce with Apache Hadoop for implementing the Netting and Exposure calculation.
- Refactoring of enterprise grid framework developed previsouly (see previous work) : implementation of event based single threaded proactor server using boost Asio.
- Integration of a Coherence data caching solution prototype + creation of a simple API. Cooperation with Oracle team on development of Coherence Incubator patterns in C++.
- Improvement of grid abstraction layer API (see previous job) :
o Asynchronous submission
o Data transfer optimization and data references
o Condition services chaining
o Status checking / cancellation for grid services using Admin API.
o Integration with DS Speedlink
- Help with on-boarding of new FX-OPTIONS products to the grid.
- Help on integration, tuning and design architecture of a Excel service on Datasynapse grid.
- Self-study - training on Haskell.
- Landsbanki - Iceland (1 week) : Setup of planning and proof of concept for the selection of a grid solution.
- Standard Chartered – London (3 months) : FX-EXOTICS – Integration of Target Profit Forward and Monte-carlo simulation blotter on the grid
o Design and development of a greenfield enterprise grid solution containing the following components : Abstraction layer API which provide high and low level access to Datasynapse routines, enterprise server based on socket adaptor optimized for high performance transfer of data, client gateway with corresponding socket client. All the developments have been done in C++ on Windows and solaris.
- KBC – Brussels (1 month) : FX-OPTIONS – Integration of Target Profit Forward product on the grid
o Design and development of an enterprise grid solution based on 3 tier : Client interface for murex flex front end client, multi-thread soap server and grid server.
o Help to the quants team for the refactoring of the pricing code in client-server with serialization of transferred data and parallelisation and integration in the grid.
o Studies for integration of Dynamic parallel version of Mersen Twister algorithm in the boost random library.
o Integration with the monte-carlo pricer in the client and engine side of the grid. Development of an abstraction layer API for Datasynapse and Platform Symphony.
Project manager, developer – Creation of a unit selection text to speech synthesizer in persian.
PACTOS
octobre 2006 - mai 2007
- environment : Linux, C++, C, Lisp
Études et formations
DEA in computer engineering -
P-M CURIE University, Paris VI
1990
Training : SPHERE-COMMUNICATION Creation of a multimedia editor on NEXT workstation
1990
Training : HAY-SPACE (member of Aerospatial and Hay Management) • conception of a multimedia author system • Training of UNIX/C (Toulouse Aérospatial).
1988
Computer degree - Paris VI
1987
DEUG A (SSM) - Paris VI
1986
Langues
Anglais - B2
Autres compétences
1983 : BAC C (equivalent) - Razi French school - Tehran
LANGUAGES
French : fluent
English : used daily
Persian : fluent