$$$
{{ $t($store.state.user.experience_value_in_dollars) }}
Senior
{{ $t($store.state.user.experience_search_name) }}
0
jobs
Data scientist
Sho Okamoto
,
London, United Kingdom
Experience
Other titles
Skills
I'm offering
Strong technical abilities with long history of delivering analytical projects to a wide variety of clients from time both as an economist and data scientist.
Able to pick up new technologies quickly and able to deliver results quickly and efficiently.
Recent projects include:
- lead on delivering a POC for public sector predicting infringements for a large Government department
- created application to swap faces and clothes of models to users
- created video analytics tool to identify rugby events from tv feeds using convolutional neural nets.
Able to pick up new technologies quickly and able to deliver results quickly and efficiently.
Recent projects include:
- lead on delivering a POC for public sector predicting infringements for a large Government department
- created application to swap faces and clothes of models to users
- created video analytics tool to identify rugby events from tv feeds using convolutional neural nets.
Markets
United Kingdom
Language
English
Fluently
Ready for
Larger project
Ongoing relation / part-time
Full time contractor
Available
My experience
2015 - ?
job
Data Scientist
Capgemini.
Notable Projects
I Lead a Proof of Concept to demonstrate the value of AI to influence fisheries policy with a large Government Department
● Lead a team of 4 data scientists and data engineers to tailor a machine learning project to assess whether some of their processes could be automated, streamlined or otherwise benefit from intelligent analysis of their data.
● Worked closely with client stakeholders to ensure that we continually understood their requirements and regularly communicated our progress and raised any concerns as early as possible.
● Worked with platform engineers to create an environment where we could conduct our analysis in which the data can be accessed easily and securely by the team. This involved using 3 AWS cloud environments with Docker containers running on each, with a dedicated machine to maintain the database.
● Engineered features from the raw data that can be fed into the machine learning analysis
● Mentored team members in engineering features and cleaning the data
● Developed the AI pipeline that was flexible, accessible, and extensible that allowed the machine learning analysis to be conducted.
● Presented results to clients and internal stakeholders.
I conducted video analysis for events detection in rugby 7s
● The requirement was to automatically tag rugby events (tackles, scrums, line outs, line breaks, tries, passes, and catches) from the videos of Rugby 7s televised feeds.
● Tagged 3000 rugby match frames to identify the various events.
● Created an environment using a GPU accelerated AWS cloud environment in order to train a YOLO v3 model to detect events from individual frames.
● Additionally created a Convolutional Neural Network model in TensorFlow to create a model to detect wider scenes and calculate the number of players on the screen.
● Used the outputs from the Machine learning models to analyse events from frames to tag events across multiple frames
● Present the method and results in an accessible way to colleagues that did not need to know the precise technicalities of the models.
● Ensure that the output was clear and intuitive without being overly technical.
I developed and maintained a machine learning framework used by 50 data scientists for a large Government Department
● As the number of projects within the department grew, there was value in creating a common framework that managed data access, data cleansing, feature engineering, machine learning, plotting charts, and assessing outputs.
● I created a fully unit tested framework with configurable parameters, and configurable modules to accommodate the varying needs of the disparate projects.
● The user interface of the framework had to be accessible to a group of data scientists of mixed levels of experiences, and more flexible for those with higher skills.
I developed a matching algorithm to match SQL data schemas
● The requirement was to automate the migration of data from one schema to another (eg the same data but from different systems).
● The algorithm I developed was used to assess whether a table from one schema was the same as another, and if not which columns overlapped.
● The algorithm used a combination of ngrams, levenshtein differences, and semantic analysis.
● Worked in an agile environment, and continually communicated my progress to software engineers and the scrum master to ensure that we were all aligned and no effort was duplicated.
I Lead a Proof of Concept to demonstrate the value of AI to influence fisheries policy with a large Government Department
● Lead a team of 4 data scientists and data engineers to tailor a machine learning project to assess whether some of their processes could be automated, streamlined or otherwise benefit from intelligent analysis of their data.
● Worked closely with client stakeholders to ensure that we continually understood their requirements and regularly communicated our progress and raised any concerns as early as possible.
● Worked with platform engineers to create an environment where we could conduct our analysis in which the data can be accessed easily and securely by the team. This involved using 3 AWS cloud environments with Docker containers running on each, with a dedicated machine to maintain the database.
● Engineered features from the raw data that can be fed into the machine learning analysis
● Mentored team members in engineering features and cleaning the data
● Developed the AI pipeline that was flexible, accessible, and extensible that allowed the machine learning analysis to be conducted.
● Presented results to clients and internal stakeholders.
I conducted video analysis for events detection in rugby 7s
● The requirement was to automatically tag rugby events (tackles, scrums, line outs, line breaks, tries, passes, and catches) from the videos of Rugby 7s televised feeds.
● Tagged 3000 rugby match frames to identify the various events.
● Created an environment using a GPU accelerated AWS cloud environment in order to train a YOLO v3 model to detect events from individual frames.
● Additionally created a Convolutional Neural Network model in TensorFlow to create a model to detect wider scenes and calculate the number of players on the screen.
● Used the outputs from the Machine learning models to analyse events from frames to tag events across multiple frames
● Present the method and results in an accessible way to colleagues that did not need to know the precise technicalities of the models.
● Ensure that the output was clear and intuitive without being overly technical.
I developed and maintained a machine learning framework used by 50 data scientists for a large Government Department
● As the number of projects within the department grew, there was value in creating a common framework that managed data access, data cleansing, feature engineering, machine learning, plotting charts, and assessing outputs.
● I created a fully unit tested framework with configurable parameters, and configurable modules to accommodate the varying needs of the disparate projects.
● The user interface of the framework had to be accessible to a group of data scientists of mixed levels of experiences, and more flexible for those with higher skills.
I developed a matching algorithm to match SQL data schemas
● The requirement was to automate the migration of data from one schema to another (eg the same data but from different systems).
● The algorithm I developed was used to assess whether a table from one schema was the same as another, and if not which columns overlapped.
● The algorithm used a combination of ngrams, levenshtein differences, and semantic analysis.
● Worked in an agile environment, and continually communicated my progress to software engineers and the scrum master to ensure that we were all aligned and no effort was duplicated.
SoMe, Processes, Framework, Feature, Convolutional Neural Network, Software, Ai, Engineering, Network, Tensorflow, Sql, Cloud, Database, Agile, Scrum master, AWS, Video, Docker, Machine learning, Scrum
2014 - 2015
job
Economist
unknown.
Notable Projects
I was involved in a project assessing whether consumers were taking unaffordable loans
● The aim of this analysis was to assess whether people were taking on credit that they could not afford. In order to assess this, we took a sample of consumers who took out a loan and those that did not, and compared their outcomes after a few months. This analysis was enabled by a statistical method called Regression Discontinuity Design (RDD).
● We collected all transaction data from 20 consumer credit firms from a period of 18 months.
● I was required to implement advanced statistical methods in R and Stata, and manage and maintain multiple related datasets.
I was involved in a project assessing whether consumers were taking unaffordable loans
● The aim of this analysis was to assess whether people were taking on credit that they could not afford. In order to assess this, we took a sample of consumers who took out a loan and those that did not, and compared their outcomes after a few months. This analysis was enabled by a statistical method called Regression Discontinuity Design (RDD).
● We collected all transaction data from 20 consumer credit firms from a period of 18 months.
● I was required to implement advanced statistical methods in R and Stata, and manage and maintain multiple related datasets.
Design, R, Stata
2010 - 2014
job
Economist
Frontier Economics.
Notable Projects
I was the lead modeller for the supply side analysis for a high profile project (the FCA payday loans price caps)
● I managed a team of analysts hired across several consultancies to assess the supply side response of imposing a price for payday loan companies.
● Using transaction data gathered from firms, we modelled the profitability of existing firms under various hypothetical price cap structures that were set by the FCA policy makers
● I was required to implement the model in an extensible and adaptable manner as the requirements for the outputs, the inputs, and the structure of the cap were continuously evolving as the analysis progressed.
I was the lead analysis assessing whether LIBOR was manipulated
● The purpose of this project was to analyse whether financial traders were attempting to manipulate the LIBOR index by analysing their submissions and their trades.
● This involved collecting and amalgamating a huge dataset that included all trades conducted over 2 years.
● I used econometric methods such as Granger Causality tests to assess whether traders' behaviours were influenced by other traders.
I was the lead modeller for the supply side analysis for a high profile project (the FCA payday loans price caps)
● I managed a team of analysts hired across several consultancies to assess the supply side response of imposing a price for payday loan companies.
● Using transaction data gathered from firms, we modelled the profitability of existing firms under various hypothetical price cap structures that were set by the FCA policy makers
● I was required to implement the model in an extensible and adaptable manner as the requirements for the outputs, the inputs, and the structure of the cap were continuously evolving as the analysis progressed.
I was the lead analysis assessing whether LIBOR was manipulated
● The purpose of this project was to analyse whether financial traders were attempting to manipulate the LIBOR index by analysing their submissions and their trades.
● This involved collecting and amalgamating a huge dataset that included all trades conducted over 2 years.
● I used econometric methods such as Granger Causality tests to assess whether traders' behaviours were influenced by other traders.
My education
University College London
MSc, Economics
MSc, Economics
Queens College University of Cambridge
Bachelors, Economics
Bachelors, Economics
n/a
Second Class Degree, N/a
Second Class Degree, N/a
Sho's reviews
Sho has not received any reviews on Worksome.
Contact Sho Okamoto
Worksome removes the expensive intermediaries and gives you direct contact with relevant talent.
Create a login and get the opportunity to write to Sho directly in Worksome.
are ready to help you
and get specific bids from skilled talent in Denmark