AI Prompt Engineer

Join an Rapidly-Growing Consumer Intelligence Platform Delivering Insights for Leading Global Brands

Company Background

We're an AI-powered consumer intelligence platform that transforms billions of data points (Google searches, social conversations, product reviews, and videos) into strategic insights for leading global brands such as Coca-Cola, Bayer, and Unilever. We deliver in days what traditional research takes months to produce.

The Role

You will define prompt engineering across the company: prompt design standards, reusable templates, evaluation, and release governance. You’ll take prompt ownership off the data science team (so they focus on modeling) and enable client-facing teams to generate consistent analysis and narratives from structured metrics and verbatims.

This role partners daily with Data Science, Data Analysts, Client Success, and Platform Engineering.

What You’ll Do

• Audit and refactor existing NLP/GenAI prompt libraries (sentiment, emotion, classification, driver extraction, summarization, etc.) and bring them to acceptance-test quality (schema compliance, accuracy targets, safety checks).

• Build a modular prompt template system (parameterized by market/segment/language/output format) used across data science and client-facing reporting workflows.

• Optimize prompts across at least 2 LLM providers (OpenAI / Claude / Gemini), document provider-specific behaviors, and maintain output consistency requirements.

• Define prompt reliability behaviors for invalid inputs, missing data, and formatting failures (validation rules, refusal rules, confidence flags, and fallback requirements for calling services).

• Implement a prompt QA program: regression test sets, automated checks (schema/consistency/safety), and a human review rubric with scorecards tracked over time.

• Create analytics + storytelling prompts that convert structured metrics/tables (e.g., brand equity, benchmarks, trends) into repeatable, executive-ready narratives with citations/evidence selection.

• Deliver enablement: publish playbooks, run prompt review sessions, and train analysts/client success on using templates and producing consistent outputs.

• Own the prompt repository and lifecycle (GitHub PR workflow, versioning, tagging, approvals, release notes, monitoring, and retirement).

Required Qualifications

Experience

• 3+ years building prompts for production LLM workflows (prompt engineering / LLM applications / applied NLP with LLMs), including ownership of prompt changes and outcomes.

• Multi-LLM: Hands-on with at least 2 providers (OpenAI GPT series, Anthropic Claude, Google Gemini); ability to explain and handle provider behavior differences.

• Prompt Engineering (Proficient): Structured outputs (JSON/tables), reusable templates, few-shot/zero-shot patterns, and multi-step prompting/tool use where applicable.

• Evaluation (Working+): Automated prompt checks + regression approach + human rubric; examples of metrics/scorecards.

• Multilingual (Working+): English-authored prompts that reliably process non-English text; can describe languages supported and evaluation method.

• Python (Proficient): 2+ years professional Python; ability to build tooling for prompt testing/evaluation.

• Orchestration (Working+): LangChain, LlamaIndex, or equivalent orchestration framework experience.

• Version Control (Working+): Git/GitHub PR workflow experience for shared libraries.

• Enablement: Evidence of documentation and training delivered to non-technical stakeholders (playbooks, workshops, recurring review cadence).

• Education: Bachelor’s in a related quantitative field or equivalent experience (4+ additional years relevant work in lieu of degree).

Preferred Qualifications

• Prompt security (prompt injection resistance, jailbreak mitigation, red teaming practices).

• CI/CD for prompts (GitHub Actions or equivalent) running regression suites on PRs and nightly.

• Azure + Databricks/Delta Lake/Unity Catalog familiarity.

• SQL skills for inspecting and validating the data behind prompts.

• Consumer insights / market research / CPG analytics background.

• Fine-tuning knowledge (tradeoffs vs prompt-only approaches).

• NLP tooling: Hugging Face, spaCy, NLTK; PySpark for data inspection.

Location & Work Arrangement

• Timezone overlap: Candidates should be able to overlap at least 4 hours with US Eastern and/or European business hours (team-dependent).

• Travel: Not required unless otherwise stated for team offsites (≤2x/year).

What We Offer

• Competitive salary

• Remote-first role

• Clear deliverables and review cadence with outcomes-based evaluation

• Opportunity to own a company-wide prompt platform: templates, evaluation, governance, and enablement.