Faisal Malik

• Design an end-to-end machine learning flow to perform semantic segmentation on biospecimens from data labeling to model serving in the AWS ecosystem
• Provision serverless infrastructure on AWS using Lambda, SQS, DynamoDB, S3, ECS, and Fargate to enable tile-level model inference parallelization.
• Automate experimentations, training, and hyperparameter tuning process using GitHub Actions dispatch workflow that provisions on-demand runners on EC2, which execute the job.
• Implement various image transformations to improve segmentation results.

• Migrated all async calls implementation to the OpenAI API to the OpenAl Batch API in the Kubeflow Pipeline, which reduced OpenAI token consumption costs by 50% and eliminated all parallel invocation instances from Kubeflow components.
• Standardize the pipeline using multiple reusable Kubeflow components, allowing all team members to productionize their Pipelines seamlessly.
• Set up CI/CD Pipeline to test and deploy from GitHub to the GCP ecosystem.
• Set up GitHub Actions Workflow Dispatch to submit and schedule Vertex Al Pipelines.
• Provisioned and managed GCP infrastructure using lac tools like Pulumi.
• Refactored experimental codes from data scientists to meet production standards and deployment.
• Integrated different services and components built by other team members so the system can run smoothly and efficiently.

All-in-One Decentralized Physical Infrastructure (DePIN) manager.

• Design and implement a Data Ecosystem in GCP from the ground up using Terraform and GitHub Actions.
• Design and implement log centralization from end-user applications to Cloud Logging.
• The logs are routed hourly to Cloud Storage and transformed upon arrival using Cloud Run Jobs with an adjusted instance size based on the log size.
• The transformed logs are stored in BigQuery for further analytics, which utilize a Materialized View to optimize Aggregate retrieval.
• This strategy enables flexible analytics to accommodate new use cases without requiring changes to the log source.
• Develop a Blockchain indexing framework using a subgraph to index multiple DePINs' on-chain earnings data.
• This allows the data to be accessed using GraphQL and analytics purposes.
• Enable self-serve analytics throughout the company by provisioning Metabase and connecting BigQuery as a data source with business-friendly data models.
• This makes the company data-driven even without a data analyst to translate their questions into SQL queries.
• Visualize core business metrics in Grafana with optimized performance and views.