Oryan Omer
Principal Software Engineer 👨🏻💻 | MLOps Engineer 🤖 | ML Engineer 🤖 | Cyber Security Specialist | Wave Surfer 🌊
Principal Software Engineer with a strong expertise in machine learning, MLOps, and Cloud Security.
With 10 years of experience developing services for both production and development environments, I have a proven track record of delivering high-quality software solutions.
Throughout my career, I have led teams, mentored junior and senior developers, improved R&D processes, designed complex architecture solutions, and managed large-scale projects.
Additionally, I have been a speaker at several conferences such as ODSC West(SF) and Europe(London) in 2022, Google Next 24 in Las Vegas, and have also participated in various webinars.
I frequently publish content on Linkedin and Medium that talks about ML engineering, MLOps, Software engineering, and tech in general.
I have a B.sc computer science from the College of Management.
In my free time, I'm surfing 🌊, snowboarding 🏂 and doing a lot of sports (:
Activities
In the following list, you will find some events and webinars where I have given a talk
API Security with GCP Service Extensions
Google Cloud Next'24, Las Vegas, April,2024
During this presentation, I will demonstrate how to use the Google Service Extension Beta feature to inspect all HTTP traffic data from the Google Application Load Balancer.
We will cover how to identify and find HTTP attacks, and then send the insights gained to Palo Alto Prisma Cloud.
Additionally, I will show a real demo use case of a development called "panw apisec sensor" that inspects all the HTTP traffic data of the load balancer and learns from it.
Session Outline:
- Overview on API Security
- Why API Security is important
- How to use GCP Service extension for API Security
- 4 levels for API Security
Data-driven retraining with production ML insights
ODSC West, San Fransisco, Nov,2022
In this talk, I show case, through ML monitoring and notebooks, how data scientists and ML engineers can leverage ML monitoring to find the best data and retraining strategy mix to resolve machine learning performance issues. This data-driven, production-first approach enables more thoughtful retraining selections, shorter and leaner retraining cycles, and can be integrated into MLOps CI/CD pipelines for continuous model retraining upon anomaly detection
Session Outline:
- Retraining groups and temporal similarity
- Drifted features and pre-preprocessing
- Drifted segments and model split
- Pipeline anomaly exclusion.
Automating ML pipelines
Data science Salon, Virtual Event, June,2022
As teams look to build and deploy models into production, they need tools that can adequately scale with them. In particular, the tools they need must allow them to quickly monitor, segment, retrain, and experiment on the data.
Session Outline:
- What is MLOps and why is data critical for it./
- How to architect a scalable and automated platform.
- Why your team should adopt a Production-First data approach.
Data-driven retraining with production ML insights
ODSC Europe, London, May 2022
It’s practically dogma today that a model's best day in production will be its first day in production. Over time model performance degrades, and thare are many variables that can cause decay, from real-world behavior changes to data drifts.
When models misbehave, we often turn to retraining to fix the problem, but is the most recent data the best data to resolve our model performance issues and get it back on track? We all acknowledge the need for data-driven machine learning monitoring that pinpoints anomalies and uncovers their root cause so we can resolve issues quickly before they impact the business.
When it comes to resolution through retraining, data selection and the retraining strategy selected are less than data-driven. Today when faced with retraining, many data teams simply select the last month or two of data to retrain on and hope that fresh really is best.
Blogs
A few of my published articles can be found here.
Data-driven retraining with production observability insights
“Refreshing” a model’s training observations isn't always enough to predict future results better. Using observability of model production data gives us a clear path to better decisions on our retraining strategies.
So you want to be API-first?
Building a great, scalable, and easy-to-integrate product starts with design. At superwise, we understand the importance of being an API-first company. Read my article and follow superwise’s journey to API-first and what considerations you should make before embarking on that journey.
Testing Data-driven Microservices
Testing microservices that work on a massive data scale is a complex task. I designed an efficient architecture to make the testing phase much easier for writing, debugging and ensuring the application integrity (Hint: It leverages Jupyter Notebook and Papermill)
Use FastAPI for Better Machine Learning in Production
Deploying ML for serving in production sounds like a complicated task, which costs too much times for developers. Check out my article to understand why you should start using FastAPI to speed up this process and save a lot of time.
Scaling data-driven microservices
Scaling data-driven microservices need to be supported by an infrastructure that can accommodate resource extension or decrease resources, but more importantly, codes should be written to support scalability
CONTACT ME TO GET STARTED
