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.