Prompt Engineer Intern

Profile We are hiring a world-class Prompt Engineer, someone with extreme clarity of thought, deep reasoning skills, and hands-on mastery of modern LLM behavior. It is for someone who understands prompting as a science: evaluation, ablation, latency tuning, hallucination control, guardrails and AGI-aligned orchestration. Requirements • Exceptional reasoning, logic structuring and analytical thinking. • Ability to design deterministic, controlled, and testable prompts for production behaviour. • Deep understanding of temperature, top-p, top-k, seed, max-tokens, repetition penalty, and their effects on model behaviour. • Strong mastery of LLM guardrails, prompt-injection defense, role separation, Meta Logic, and system level prompt design. • Solid understanding of modern LLMs and NLP systems (GPT, Claude, LLaMA, Qwen, Mistral, etc.). • Knowledge of low-latency model families, quantization (GGUF, INT4/8), and inference trade-offs. • Clear understanding of RAG, embeddings, vector DBs, including when not to use them. • Ability to run evaluation pipelines (accuracy, F1, hallucination rate, consistency, JSON validity). • Highly detail-oriented, structured thinker, fast learner. • Strong communication, able to explain reasoning clearly and concisely. • Comfortable building measurable experiments instead of intuition-based prompting. Responsibilities • Design high-accuracy prompts for reasoning, extraction, classification, and multi-step workflows. • Build robust system + user prompt architectures with strict constraints and safety layers. • Run temperature, seed, sampling, and ordering ablations to achieve deterministic and stable performance. • Optimize prompts for latency, consistency, cost, and token efficiency. • Reduce hallucinations using verification loops, guardrails, retrieval control, and role conditioning. • Architect governing instructions for multi-agent systems and AGI-aligned assistants. • Build hierarchical prompt layers (system → meta-controller → task-level → tools). • Select models based on latency, throughput, context window, and domain-specific performance. • Collaborate with AI engineers on quantization, distillation, caching, and runtime optimizations. • Maintain versioned prompt libraries with clear documentation, reproducibility, and experiment tracking. • Work closely with ML, product, and engineering teams to deliver production-grade, scalable prompt flows. • Write clear technical documents: architecture diagrams, ablation summaries, evaluation reports, and runbooks. What You’ll Gain • Deep exposure to real AGI workflows, multi-agent prompting, and large enterprise systems. • Mastery in experimental prompting, evaluation science, and LLM orchestration. • Opportunity to directly influence global flagship AI products across industries. • Mentorship and collaboration with top engineers, AI researchers, and product innovators.