Senior Machine Learning Engineer - NLP

📋 Description

• • Shape the next generation of AI-native customer engagement. As a Senior Machine Learning Engineer – NLP at Level AI, you will architect and deploy production-grade Large Language Model (LLM) systems that turn millions of contact-center conversations into real-time strategic intelligence for Fortune-500 clients.
• • Own the end-to-end lifecycle of LLM-powered agents. From ideation and data curation to prompt engineering, retrieval-augmented generation (RAG), fine-tuning, and scalable deployment on Kubernetes/GPU clusters, you will deliver features that agents and supervisors rely on every minute of every day.
• • Pioneer reasoning, planning, and memory modules that let agents autonomously resolve complex customer issues, escalate only when necessary, and continuously learn from outcomes. Your work will be the backbone of products that reduce average handle time by double-digit percentages while boosting CSAT scores.
• • Experiment fearlessly with the open-source and frontier-model ecosystem—LLaMA, Mistral, Claude, GPT-4/4o—and benchmark them for latency, cost, and accuracy under strict enterprise SLAs. You will publish internal white-papers that steer our quarterly model-adoption roadmap.
• • Collaborate daily with product managers, full-stack engineers, and UX designers in a fast-moving, product-led environment. You will translate ambiguous business requirements (“make the AI sound more empathetic”) into measurable objectives and reproducible experiments.
• • Champion MLOps best practices: CI/CD for prompts, automated evaluation suites, drift detection, and guardrails that keep hallucinations and toxicity near zero. You will mentor junior engineers and establish coding standards that scale across squads.
• • Optimize multi-turn dialogue systems for low-latency inference (<300 ms) at 10k+ concurrent sessions. This includes GPU kernel fusion, quantization, speculative decoding, and intelligent caching strategies that cut cloud spend without compromising user experience.
• • Leverage vector search, knowledge graphs, and hybrid retrieval (dense + sparse) to ground every LLM response in authoritative, up-to-date knowledge bases. You will design ingestion pipelines that refresh embeddings within minutes of policy or product changes.
• • Contribute to our real-time analytics layer—streaming billions of events through Kafka/Flink—to feed reinforcement-learning-from-human-feedback (RLHF) loops that continuously improve agent performance.
• • Present quarterly to the executive team, translating technical breakthroughs into revenue impact: “Our new summarization model saves 40 QA hours per week, worth $1.2 M annually.”
• • Stay plugged into the global NLP research community. Whether it’s NeurIPS, ICML, or an arXiv drop at 2 a.m., you’ll rapidly prototype promising techniques and quantify ROI within days, not months.
• • Enjoy high autonomy and visibility. Your code ships to production weekly, touches millions of end-users, and is showcased by customers on earnings calls as a competitive differentiator.