Preethi Kumar

Skills / Technologies / Tool: Python, Kubernetes, GraphQL, Rust, REST API, GitHub, GitHub actions, AWS, GCP, Docker, Airflow.

• Responsibilities:
• Lead Python SDK development, setting standards and providing guidance on style and design patterns.
• Envision user experience in different personas (Data scientist, ML Engineer, MLOps, etc) and advocate for better API that satisfies these personas.
• Continuously look for opportunities to optimize performance by statistically profiling and benchmarking.
• Improved AI model inference times by 30% by introducing concurrency in request handling.
• Developed and deployed CI/CD tooling using GitHub actions, tailoring processes to optimize the development workflow.
• Lead all technical and architectural decision making. Plan long term technical roadmap.
• Evaluate new technologies and tools, and demonstrate proof of value to the team. Mentor on how to use and integrate it with the platform.
• Designed and built robust API endpoints to package and deploy AI/ML models across various cloud environments, on prem or edge locations demonstrating ability to work with complex distributed systems.
• Designed and built FastAPI based microservice that packages and serves user's python models.

Skills / Technologies / Tool: Python, Django, Docker, Jenkins, AWS, Postgresql

• Designed, developed and test application module solutions using Python and Django.
• Selected and implemented modern design patterns to update legacy code bases, improving functionality and cloud infrastructure utilization
• Collaborated with a cross-functional team to redesign existing modules, resulting in a 15% increase in user engagement within 1.5 months."

• Skills / Technologies / Tool: Python, Flask, Airflow, Spark, Tableau, Neo4j, GraphQL, Julia, Java, JavaScript, TypeScript, React, NodeJS, D3, JupyterLab, Docker, OpenStack, Postgresql, Google Cloud Platform.
• Developed end-to-end data pipelines, selecting and tailoring tools like Airflow and Spark to create scalable solutions
• Implemented and evaluated machine learning algorithms and neural networks for diverse industries.
• Worked on ETL (Extract, Transform, Load) tasks, maintaining data integrity and verifying pipeline stability.
• Evaluated and deployed new technologies and tools, demonstrating their value to the team and integrating them into existing processes.
• Present at technical conferences, community sessions and internally.

• Skills / Technologies / Tools: Ruby, Python, Git, Linux, chef, Puppet, Ansible, Jenkins, Docker, Postgresql, Mysql, OpenStack.
• Designed and built infrastructure automation tools, selecting and tailoring technologies like Chef, Puppet, and Ansible to deploy next-generation platforms. In one the projects, this reduced server provisioning time by 60% and decreased configuration errors by 40%.
• Developed and deployed monitoring and management processes for OpenStack cloud infrastructure services, ensuring high availability and efficient scaling.
• Provide 24/7 on-call infrastructure support to maintain high availability.

Purpose of this position was to set up a Emergency reporting and notification website which works by crowd sourcing information via reports submitted through various applications like Twitter, Email, SMS, online forms and RSS feeds. It uses a platform called Ushahidi, an open source crowdsourcing application.

Skills /Technologies / Tools: Java, XML, SQL, PL-SQL, HTML, CSS, Python, EJB 1.x/2.0, JNDI, JDBC, Servlets, JSP., Eclipse, Oracle.

Responsibilities:

• Have performed all the software development life cycle processes including requirements gathering, analysis and design, development, release and version control, contemporary testing methodologies and deployment management.
• Took several initiatives to deploy best practices, such as team programming, test-driven development, continuous quality improvement reducing post-release bug fixes by 50%.
• Fostered stronger client relationships through focusing on their business objectives rather than just functional requirements, bridged the gap in requirement collection, and yielded a 97% error-free product.