Jobs at White Circle

Train, fine-tune, and deploy large-scale multimodal models (vision, video, audio, speech). Build multimodal data pipelines, run experiments, optimize MoE and inference (quantization, distillation, low-latency), and develop multimodal alignment/RLHF pipelines. Define evaluation benchmarks and ship models to production.

Train and post-train LLMs using SFT/RLHF/DPO, build reward models and large-scale data pipelines, run distributed multi-GPU training, design evaluation benchmarks, and optimize models for production inference and serving.

Build and maintain scalable ML infrastructure for LLM post-training, RL, evaluation, inference, and agentic workflows. Design data-control systems, optimize training and inference performance across networking, storage, and compute, and deliver tooling for experiments, evals, dashboards, reproducibility, and failure inspection. Collaborate with researchers and product to iterate models and infrastructure.

Build and automate GTM systems to acquire and retain customers: own CRM, design GTM data model, create pipelines, deploy AI agents, build integrations and enrichment workflows, and iterate on outbound/inbound playbooks to generate qualified pipeline.

Design and run adversarial single- and multi-agent environments to find concrete failure modes in LLM agents. Orchestrate large-scale experiments against external APIs and internal models, instrument emergent behaviors, catalogue failures, and translate findings into internal models and public research.

Owner of model-to-production infrastructure: integrate models into serving paths, build rollout pipelines, implement smoke/quality/latency gates, run canary/A-B rollouts, operate Kubernetes GPU deployments, create observability dashboards, debug full-stack production issues, and optimize serving cost and reliability.

Design and run empirical experiments to find how LLM agents fail in realistic settings; build automated audit agents, fine-tune and run models, perform white-box and black-box investigations, produce evals and publish findings that feed into product guardrails.

Build, own, and maintain internal benchmark suites for single- and multi-turn content and agentic guardrails; create benchmarks distinguishing model capabilities; extend evals to new verticals and product features; collaborate on research quantifying realistic LLM and agent failure modes; support product integration of core model functionality.