Ascra Technologies - CUDA Developer - C/C++ Programming (5-8 yrs)

Position Overview :

We are seeking an experienced CUDA Developer with a strong background in parallel computing and performance optimization.

The ideal candidate will have a proven track record of solving complex performance issues and optimizing CUDA-based applications for efficiency.

This role involves working on cutting-edge technologies and providing solutions that can scale and perform efficiently.

If you're passionate about high-performance computing and enjoy tackling complex algorithm optimization challenges, we'd love to hear from you.

Key Responsibilities

Performance Analysis and Optimization:

Identify and resolve performance bottlenecks in CUDA applications to improve efficiency and resource usage.

Analyze and optimize algorithms for parallel processing, leveraging CUDA's capabilities to improve overall execution time and system throughput.

Development and Coding :

- Develop high-performance applications using CUDA (version 12.0 or above) to leverage GPU computing for improved performance.

- Use C/C++ programming skills to write efficient and maintainable code.

Continuous Improvement and Research :

- Stay updated with the latest advancements in CUDA technologies, parallel computing, and performance optimization best practices.

- Propose and implement improvements to codebase and algorithms based on new techniques, tools, or frameworks in the field of CUDA development.

Cross-Functional Collaboration :

- Work closely with product teams, system architects, and other developers to identify areas for performance improvements.

- Provide expertise and solutions to other teams, helping them overcome challenges related to performance and CUDA-specific optimizations.

Testing and Debugging :

- Perform unit and integration tests to ensure the accuracy and efficiency of CUDA applications.

- Debug and troubleshoot complex issues related to GPU resource management and memory handling.

Required Qualifications

Educational Background:

- Bachelor's degree in Computer Science, Computer Engineering, or related fields.

- Master's degree is a plus, but not required.

Experience :

- 5+ years of software development experience.

- At least 2 years specifically in CUDA development with a focus on performance optimization.

- Experience working with Large Language Models (LLM) or similar high-performance computing applications is a plus.

Problem-Solving and Collaboration:

- Strong ability to solve complex problems related to parallel computing and algorithm optimization.

- Ability to work independently as well as part of a team, providing technical expertise and collaborating effectively with cross-functional teams.

Communication :

- Excellent communication skills to explain complex issues and solutions clearly to non-technical team members.

- Ability to document technical processes and provide insightful analysis.

Performance Optimization :

- Experience with CUDA Performance Optimization (e.g , memory management, efficient kernel launch configurations, minimizing memory latency).

- Experience in using profiling tools such as NVIDIA Nsight, CUDA Profiler, or other similar tools for performance tuning.

LLM and High-Performance Computing (HPC) :

- Knowledge of working with Large Language Models (LLM) and other high-performance computing systems.

- Familiarity with AI/ML model acceleration using CUDA for high-performance tasks.

Software Development Practices :

- Experience with version control systems (e.g, Git, SVN).

- Familiarity with Agile development practices and working in a collaborative, fast-paced environment.

- Strong debugging skills, with the ability to identify and resolve performance and concurrency issues efficiently.

Optional Skills :

- Experience with machine learning frameworks such as TensorFlow, PyTorch, or similar.

- Familiarity with cloud computing platforms (AWS, GCP, Azure) for scalable GPU resources.

- Knowledge of data structures and algorithms optimized for parallel processing environments