AI Training and Prompt Engineering Specialist

This is a fully remote position, open to applicants in Poland.

📋 Description

• Take ownership of YOlanda’s system prompt, which includes defining the persona, tone, safety protocols, refusal behaviors, and prompts for user interactions.

• Establish what is included in YOlanda’s domain and what should be transferred to YOnC.

• Ensure seamless transitions during conversations, maintaining a consistent voice.

• Develop the few-shot library and oversee context management.

• Create evaluation sets that encompass test cases, pass criteria, and trackable scores across various releases.

• Assess and critique outputs utilizing Reinforcement Learning from Human Feedback (RLHF).

• Transform subpar responses into reference and fine-tuning data.

• Verify pricing plans, balances, and YOYO$ calculations, aiming for zero inaccuracies in financial information.

• Conduct red-team exercises on adversarial prompts, prompt injections, jailbreaks, and misuse of top-up and reward systems.

• Write Python scripts to execute and evaluate assessments at scale.

• Collaborate with APIs, including streaming, function calls, and structured outputs.

• Maintain prompt versions in Git.

• Create pull requests for the LangChain and LangGraph builds in partnership with the two engineers responsible.

• Engage with the model daily without the responsibility of managing other team members.

⛳️ Requirements

• Minimum of 2 years of experience working with large language models in a production environment.

• Proficient in Python for scripting, API client development, data management, and testing.

• Familiarity with LangChain and LangGraph, with the capability to read and modify existing graphs without causing disruptions.

• Experience in production retrieval processes, including chunking, embeddings, reranking, and grounding.

• Outstanding written communication skills in English.

• Bilingual proficiency in Spanish and English at an equivalent level.

• Systematic approach to experimentation: adjust one variable, test, document results, and proceed.

• Preferred: Experience in fine-tuning (SFT, DPO, LoRA).

• Preferred: Knowledge of evaluation frameworks (promptfoo, Braintrust, LangSmith, DeepEval).

• Preferred: Background in mobile, telecommunications, or fintech sectors.

• Preferred: Training in editorial practices or linguistics.

🏝️ Benefits

• Flexible working hours.

• Provision of corporate equipment for work purposes.

• Competitive salary package.

• Genuine opportunities for personal and professional development.