How can I help?

I am an independent software engineer that specialises in data. Having spent a number of years working as a full time machine learning engineer, I now work on a contract basis. I have a masters degree in computer science with a specialisation in data mining. I enjoy working independently and as part of a team.

Please get in contact if you'd like to explore opportunities to work together.

Contact Portfolio

What can I do for you?

Datascience

AI Pipelines

Design and implement the data preparation and model training, deployment and inference pipelines.

Data management

Manage and clean your company's data, integrate external sources and expose it for business intelligence with ETL processes.

Micro services

Architecture and deploy micro services for your organization through Restful APIs.

CI/CD

Automate, speed-up and optimize the integration, delivery and deployment of your applications.

Skillset

I am a back-end systems specialist and data expert; however, I also have experience as a full stack engineer and can pick up new technologies quickly.

My mission is to make sure your organization's data is well prepared, clean and reliable. I can help you create large scale data pipelines that may involve bringing many different systems together.

As a data engineer, I can help you choose the right tools for the job, and combine them to create solutions to enable your company’s business processes with data pipelines.

Check out how I helped MyTraffic in collecting and leveraging pedestrian footfall data to predict best retail locations...
Check out the big data pipelines I developed for the BI team at Gumgum...

Languages

Frameworks
Apache Spark logo

My mission is to help you understand if and where machine learning can be useful for your company, and implement the AI models and systems for you.

As a data scientist, I can help you create data pipelines for machine learning (extraction, cleaning, augmentation) and deploy state of the art models suited to your problematic.

Check out my work on fake faces recognition on a 500GB video dataset…
Check out my work on building a bot detection engine… Check out my open-source image classification app…

Languages

Frameworks
Keras Logo

I can work with your team to shorten and optimize your application lifecycle process, and facilitate fast and clean deployments via standardization and automation. I am also able to architect your systems using a cloud provider of your choice (AWS, GCP, Azure, IBMCloud)

If you need to deploy CI/CD pipelines across your organization or expose applications as micro-services, on-premises or on the cloud, I can do it for you.

Languages
Frameworks
Docker Logo

As a ML engineer, I bridge the gap between data scientists, data and devops engineers. I can help you on the following tasks :

  • Design, architect and deploy end-to-end AI pipelines.
  • Improve model deployment, training and monitoring processes
  • Scale existing models, reduce time to market for AIs
  • Work with data scientists to increase their productivity
  • Evaluate and deploy MLOps tools (MLFlow, DVC, W&B, Kubeflow...)
ml_workflow_oreilly
Source : Introducing MLOps - O'reilly ®
Frameworks
Kubeflow Logo

Recommendations

Latest work

Here are some examples of projects I worked on, in various contexts (start-ups, bigger companies, independently, as an employee).

Contact me Resume


At MyTraffic, I implemented data science models at high scale, and deployed them on the cloud for pedestrian traffic insights.

Jan 2020
Big data pipeline at GumGum

Built and maintained batch and real-time data pipelines to provide business insights on a continuous basis. These pipelines were downstream from an ad-serving platform processing 50 TB of data a day.

Feb 2019
Cruise control architecture

Deployed a suite of monitoring and management tools across multiple Kafka clusters. These tools reduced the engineering time of certain cluster operations (scaling, rebalancing) by 90% as well as enhanced the overall visibility into the health of these clusters.

Apr 2019
Thumbnail [100%x225]

Developed a high-scale programmatic inventory forecasting web tool. This tool used an ARIMA forecasting model and hundreds of terabytes of historical data to produce a forecast of ad inventory. In close cooperation with the operational teams, we provided the end-users the ability to determine achievable ad campaigns goals with confidence using the forecast results.

Thumbnail [100%x225]

A simple computer vision system able to classify images into 1000 object categories, such as keyboard, pencil, and many animals. Visit the UI or try it out below.

Feb 2020

Deeplabv3+

DeepLab is a state-of-art deep learning model for semantic image segmentation, where the goal is to assign semantic labels (e.g., person, car, dog, bicycle and so on) to every pixel in the input image.

DeepLabv3+, extends DeepLabv3 by adding a simple yet effective decoder module to refine the segmentation results especially along object boundaries.

According to Papers With Code, it has the best accuracy and is the current state of the art on the Pascal VOC dataset.

The code is available here.

Try it out !

Or send one of the sample images...

Click to send
Click to send
Click to send