About

We’re seeking a sharp, analytical, and results-driven Data Scientist to help us improve the effectiveness of our cold email campaigns through prompt optimization. You will work at the intersection of data analysis, prompt engineering, and A/B testing, collaborating closely with our internal team to ensure we’re consistently getting the best possible results from our outreach. This is not a generic data role, we’re looking for someone who thrives in fast-paced, experimental environments, can work with marketing data, and understands how language impacts behavior. We have two scenerios: 1) We run video transcripts + prompts through AI. We want the citations of the outputs, but currently we do not include links to sources as we have not been able to get an LLM to provide accurate URLs. We want someone to run a few examples and record tokens used/price then test using Perplexity and updating the prompts so that we can achieve results with citations at no additional cost. 2) We have built an AI cold email system and want to optimise through data science and prompt/context optimisation. So reviewing the performance of prompts that are used to create emails (Subjects and Body) and create a self improving feedback loop. You Must Have: Deep expertise in prompt engineering, particularly for output control and API-based LLM use Experience with Perplexity, GPT-4, Claude, or similar models Strong statistical analysis and experimentation design skills Proven track record with email campaign optimization (open rates, CTR, CVR, etc.) Hands-on experience with Python or R for automation/scripting Familiarity with Zapier/n8n/Make or custom workflows a plus Key Outcomes: A working system where our AI cold emails continuously improve based on real performance data A set of updated prompts (or models/tools) that can return citations/sources in transcript-based outputs without increasing API costs What This Is Not: This is not a content writing or ChatGPT tinkering gig. It’s a performance-driven, data-backed technical role for someone who knows how to make prompts deliver business outcomes. To apply: 1) Take the IMA Personality Test: https://imahigh.com/ and include your Color as the first word of your proposal. 2) Include a Loom video (max 5 mins) showing how you’ve used prompt + performance data to improve AI outputs and optimize cold email systems like the two scenarios described. ⚠️ Important: Applications without these will be automatically rejected.