20 Nov
|
Spydra
|
Secunderabad
20 Nov
Spydra
Secunderabad
Positions: 1 | Experience: 4–7 years
About the role You will establish reproducible performance measurement and regression testing for inference services.
Key responsibilities
- Design benchmark suites across models, contexts, concurrency levels and request patterns.
- Measure TTFT, TPOT, inter-token latency, throughput, concurrency and GPU utilisation.
- Run load, stress, endurance and scalability tests with controlled comparisons.
- Compare GPU types, model variants, quantisation and serving frameworks.
- Build automated performance regression pipelines and observability dashboards.
- Evaluate cost per million tokens and power/energy efficiency.
Required technical skills Python, Prometheus, Grafana, NVIDIA DCGM, MLPerf methodology, GenAI-Perf or equivalent, load-testing tools, statistics.
Expected deliverables Automated benchmark suite, performance baselines, comparison reports and regression dashboards.
📌 AI Inference Performance & Benchmark Engineer (Secunderabad)
🏢 Spydra
📍 Secunderabad