Pika
ML Engineer, Inference & Optimization
$250K–$350K
US remote · Remote · Full-time · Posted Jun 23, 2026
- Experience
- 5+ years engineering experience.
About the job
Designs and optimizes inference systems and GPU workloads to deploy Pika’s video and language models efficiently at scale.
What you'll do
- Implement inference acceleration, including attention optimization and quantization.
- Optimize GPU parallelism and develop high-performance kernels and distributed workloads.
- Collaborate with research and engineering teams to deploy video-generation and large language models; contribute to training-efficiency improvements.
What they ask for
- 5+ years of engineering experience, with a track record in inference acceleration and large-scale model deployment.
- Expertise in inference optimization, including quantization, attention acceleration, and deep learning compiler stacks.
- Deep knowledge of GPU programming (CUDA, NCCL) and distributed inference parallelism (SP, TP, PP).
- Familiarity with video-generation models and large language models.
Nice to have
- Experience deploying high-throughput video or real-time streaming models.
- Familiarity with distributed training and optimization toolkits; open-source AI infrastructure or compiler contributions; or startup/prototyping experience.
Pay
$250K – $350K (currency and whether base salary or total pay not stated)
Tools
CUDA, NCCL, TP, SP, PP
AI focus
Accelerates inference and production deployment of video-generation and large language models; may also improve model training efficiency.
Model-written summary · the original posting is authoritative