Lead II - Embedded Software

UST; Bengaluru; On-site; INR 1,000,000 - 1,500,000; Full time

Job summary

UST in Bengaluru is seeking a skilled professional for benchmarking AI models, optimizing performance, and analyzing system-level metrics. This role involves working with various AI models and hardware platforms, offering insights from performance comparisons.

The ideal candidate will have experience in performance benchmarking, AI/ML framework knowledge, and programming skills in C/C++ and Python. A strong understanding of Linux systems and profiling tools is required.

Qualifications

• Strong experience in performance benchmarking and system analysis.
• Good understanding of AI/ML models and inference frameworks (PyTorch, ONNX, etc.).
• Experience with Linux systems and performance tools.
• Knowledge of CPU/GPU/NPU architectures and memory systems.
• Experience with profiling tools such as perf, VTune, and Nsight.
• Programming proficiency in C/C++, Python, or Perl.

Responsibilities

• Benchmark AI models such as OpenVLA, YOLOv8, and Whisper.
• Measure performance metrics including latency and throughput.
• Set up and run benchmarking pipelines across hardware platforms.
• Perform system-level analysis to identify performance bottlenecks.
• Optimize AI models using profiling tools.
• Integrate AI models into application pipelines and measure performance.
• Compare performance across platforms and generate insights.
• Automate benchmarking execution and data collection.
• Prepare technical reports and performance summaries.

Skills

Performance benchmarking

AI/ML models understanding

Linux systems

CPU/GPU/NPU architectures

Profiling tools

C/C++ programming

Python programming

Benchmarking

Graphics processing unit knowledge

Tools

perf

VTune

Nsight

Job description

Role Description

Benchmark AI models such as OpenVLA, YOLOv8, Whisper, Pi Droid, Groot, and MLPerf Client and similar workloads.

Measure performance metrics including latency, throughput (FPS), memory usage, and power efficiency.

Set up and run repeatable benchmarking pipelines across different hardware platforms.

Perform system‑level analysis to identify bottlenecks in CPU, GPU, memory, and data pipelines.

Optimize AI models and pipelines using profiling tools and performance tuning techniques.

Integrate AI models into application pipelines (e.g., robotics or edge AI workflows) and measure end‑to‑end performance.

Compare performance across platforms (AMD, NVIDIA, Qualcomm, etc.) and generate competitive insights.

Automate benchmark execution, data collection, and reporting.

Prepare clear technical reports and performance summaries for stakeholders.

Key Responsibilities

• Benchmark AI models such as OpenVLA, YOLOv8, Whisper, Pi Droid, Groot, and MLPerf Client and similar workloads.
• Measure performance metrics including latency, throughput (FPS), memory usage, and power efficiency.
• Set up and run repeatable benchmarking pipelines across different hardware platforms.
• Perform system‑level analysis to identify bottlenecks in CPU, GPU, memory, and data pipelines.
• Optimize AI models and pipelines using profiling tools and performance tuning techniques.
• Integrate AI models into application pipelines (e.g., robotics or edge AI workflows) and measure end‑to‑end performance.
• Compare performance across platforms (AMD, NVIDIA, Qualcomm, etc.) and generate competitive insights.
• Automate benchmark execution, data collection, and reporting.
• Prepare clear technical reports and performance summaries for stakeholders.

Required Skills

• Strong experience in performance benchmarking and system analysis.
• Good understanding of AI/ML models and inference frameworks (PyTorch, ONNX, etc.).
• Experience with Linux systems and performance tools.
• Knowledge of CPU/GPU/NPU architectures and memory systems.
• Experience with profiling tools such as perf, VTune, Nsight or similar.
• Programming proficiency in C/C++, Python/Peral.

Skills

• Embedded software development
• Graphics processing unit
• Robotics
• Artificial intelligence
• Benchmarking
• Python
• C++