AI Engineer, Gopuff

• MSc or PhD in Computer Science, Machine Learning, or equivalent research experience with significant contributions to AI/ML literature
• Established experience in large-scale machine learning research with demonstrated impact on real-world systems
• Deep expertise in transformer architectures, large language models, and modern pre- and post-training paradigms
• Mastery of advanced fine-tuning techniques including LoRA/QLoRA, adapter methods, and parameter-efficient transfer learning
• Research experience with agentic AI frameworks, multi-agent systems, and declarative programming approaches (DSPy, LangChain ecosystem)
• Strong systems engineering capabilities with PyTorch, distributed training, and cloud-native ML infrastructure
• Track record of publications in top-tier venues (NeurIPS, ICML, ICLR, AAAI) or equivalent industry impact
• Commitment to responsible AI development and alignment research

Desirable

• We are building the next generation of hyper-personal AI systems that fundamentally transform how humans interact with businesses
• Led by our Chief AI Scientist, the Personal Superintelligence Lab at Gopuff develops cutting-edge agentic AI solutions that reason about complex user contexts, preferences, and real-world constraints to deliver superhuman performance in personalized shopping assistance
• You will advance the state-of-the-art in prediction, alignment, grounding, and multi-agent orchestration while deploying these breakthroughs at massive scale
• You will own core components of our instant shopping personal intelligence—context engineering, grounding, alignment, and serving—so technology accelerates customers to their goals, leveraging our unparalleled delivery speed
• You’ll design context engineering strategies and real‑time retrieval, build SFT pipelines, and implement reinforcement learning from human feedback (RLHF) and verifiable rewards (RLVR) using GRPO/GSPO
• You’ll use relevant APIs for data augmentation and ship low‑latency, reliable inference (batching, caching, quantization, streaming) with strong observability and safety
• In close partnership with Engineering and Data Science, you’ll run experiments end‑to‑end and deliver production models that drive measurable lifts in business KPI under privacy‑by‑design and rigorous evaluation
• Advanced Context & Grounding Research:
• Pioneer novel architectures for multi-modal context integration across temporal, spatial, and behavioral dimensions
• Develop foundational techniques for real-time knowledge grounding with dynamic constraint satisfaction
• Research declarative programming paradigms (DSPy) for robust prompt compilation and systematic LLM behavior specification
• Design and develop multi-agent orchestration systems (LangChain, LangGraph, CrewAI, Autogen, LlamaIndex) that exhibit emergent reasoning capabilities
• Alignment & Learning Systems:
• Advance reasoning-centered supervised fine-tuning methodologies with novel data curation, synthetic generation, and quality assurance frameworks
• Develop and evaluate model reasoning (task decomposition, tool-use, and self-correction) to improve grounded recommendations and task completion under real-world constraints
• Research parameter-efficient adaptation techniques including LoRA variants, Text-to-LoRA synthesis, and dynamic adapter routing
• Pioneer RLHF and RLVR with emphasis on scalable oversight
• Develop next-generation policy optimization algorithms beyond GRPO/GSPO with formal safety guarantees
• Safety & Robustness:
• Research interpretability, controllability, and alignment verification for production agentic systems
• Develop formal methods for safe exploration, reward hacking prevention, and distributional robustness
• Pioneer privacy-preserving techniques and federated learning approaches for personal AI systems
• Systems & Deployment:
• Research efficient inference architectures including novel quantization, caching, and streaming paradigms
• Develop evaluation frameworks that bridge offline metrics with online performance and safety criteria
• Build foundational infrastructure for rapid experimentation and deployment of research breakthroughs