QA Engineer (Performance)- Dallas at Photon

Job description

We are looking for a Performance QA Engineer to specialize in benchmarking and optimizing our Agentic AI platform . You will be the gatekeeper of the "User Experience of Thought," ensuring that as our AI agents plan, reason, and execute tasks, they do so within acceptable timeframes and cost-efficiency boundaries. Your mission is to stress-test the entire AI pipeline—from the initial prompt to the final autonomous action—identifying bottlenecks in LLM response times, RAG (Retrieval-Augmented Generation) retrieval speeds, and third-party API orchestration. Key Responsibilities • Latency Benchmarking: Measure and optimize TTFT (Time to First Token) and Total Request Latency for complex agentic workflows that involve multiple reasoning steps. • Agentic Loop Stress Testing: Simulate high-concurrency environments to see how the system handles hundreds of autonomous agents running simultaneously, particularly focusing on API rate limits and GPU/compute bottlenecks. • RAG Performance Analysis: Test the speed and efficiency of the vector database retrieval process. Identify how increasing the "context window" size impacts overall system performance. • Token Throughput Monitoring: Analyze the "tokens per second" (TPS) metrics and identify when model-switching (e.g., from a large model to a smaller one) is necessary to maintain performance. • Cost vs. Performance Optimization: Create reports that balance performance gains against token costs, helping the team find the "sweet spot" for production-grade agents. • Orchestration Bottleneck Identification: Use profiling tools to find delays in the "hand-off" between different agents or between the agent and external tools (APIs, databases). • Automated Performance Regressions: Integrate performance testing into the CI/CD pipeline to ensure that new prompt versions or architectural changes don't degrade the agent's speed. Required Skills & Qualifications • Experience: 8+ years in Performance Engineering, with a specific focus on AI/ML applications or high-concurrency distributed systems. • Tooling Proficiency: Expert-level experience with performance testing tools like Locust, JMeter, or k6 , specifically customized for Python-based AI backends. • Python Mastery: Strong ability to write custom scripts to simulate complex, multi-step user/agent interactions. • AI Infrastructure Knowledge: Understanding of LLM-specific performance factors, such as quantization, KV caching, and the impact of different model architectures on latency. • Observability Expertise: Experience with tools like Prometheus, Grafana, LangSmith, or Weights & Biases to monitor system health and AI-specific metrics. • Database Performance: Experience testing the query latency of Vector Databases under heavy load. Compensation, Benefits and Duration Minimum Compensation: USD 38,000 Maximum Compensation: USD 133,000 Compensation is based on actual experience and qualifications of the candidate. The above is a reasonable and a good faith estimate for the role. Medical, vision, and dental benefits, 401k retirement plan, variable pay/incentives, paid time off, and paid holidays are available for full time employees. This position is not available for independent contractors No applications will be considered if received more than 120 days after the date of this post Job Description Template

Requirements

Must-haves

Latency Benchmarking: Measure and optimize TTFT (Time to First Token) and Total Request Latency for complex agentic workflows that involve multiple reasoning steps.
RAG Performance Analysis: Test the speed and efficiency of the vector database retrieval process. Identify how increasing the "context window" size impacts overall system performance.
Token Throughput Monitoring: Analyze the "tokens per second" (TPS) metrics and identify when model-switching (e.g., from a large model to a smaller one) is necessary to maintain performance.
Cost vs. Performance Optimization: Create reports that balance performance gains against token costs, helping the team find the "sweet spot" for production-grade agents.
Orchestration Bottleneck Identification: Use profiling tools to find delays in the "hand-off" between different agents or between the agent and external tools (APIs, databases).
Automated Performance Regressions: Integrate performance testing into the CI/CD pipeline to ensure that new prompt versions or architectural changes don't degrade the agent's speed.
Required Skills & Qualifications
Experience: 8+ years in Performance Engineering, with a specific focus on AI/ML applications or high-concurrency distributed systems.
Tooling Proficiency: Expert-level experience with performance testing tools like Locust, JMeter, or k6 , specifically customized for Python-based AI backends.
Python Mastery: Strong ability to write custom scripts to simulate complex, multi-step user/agent interactions.
AI Infrastructure Knowledge: Understanding of LLM-specific performance factors, such as quantization, KV caching, and the impact of different model architectures on latency.
Observability Expertise: Experience with tools like Prometheus, Grafana, LangSmith, or Weights & Biases to monitor system health and AI-specific metrics.
Database Performance: Experience testing the query latency of Vector Databases under heavy load.
Compensation, Benefits and Duration
Minimum Compensation: USD 38,000
Maximum Compensation: USD 133,000
Compensation is based on actual experience and qualifications of the candidate. The above is a reasonable and a good faith estimate for the role.
This position is not available for independent contractors
No applications will be considered if received more than 120 days after the date of this post
Job Description Template

Nice-to-haves

None flagged.

Application prep

Use the role’s language and application details while tailoring your materials.

Keywords to mirror

aillmapici/cdmlpython

Platform: oracle-cloud
Effort: medium

Benefits

Benefits detected in the listing.

Dental

Vision

401k

Paid Time Off