← Back to home

One context. One reviewed draft.

This sample shows the kind of concise draft Aletheia is designed to prepare. No sign-up. No install. If the note below sounds like something you would actually edit and send, the product follows the same review-first workflow.

Input — LinkedIn profile + your resume

Priya Raman

Senior ML Engineer at DeepMind · ex-Stripe

London, United Kingdom

ML engineer focused on inference-time efficiency. Previously built fraud models at Stripe. Recent talk on sparse attention at NeurIPS 2025.

Recent post

After two months on sparse-attention inference, the real win wasn't latency — it was the smaller models we could now deploy to edge.

Your resume snippet

Nagarjun Mallesh — MS Computer Science, Boston University. Built fraud-detection pipelines processing 4M events/day at a fintech startup. Open-source contributor: pytorch/serve. Looking for ML infra roles where inference cost matters.

Output — connection note (322 chars)

Hi Priya — read your NeurIPS talk on sparse attention. I worked on inference-cost reduction at a fintech (smaller scale, fraud pipelines) and the edge-deploy angle in your recent post matched what we saw — smaller models often unlocked more than raw latency. Would love to follow your work.

Generated in 7.4s · sanitised · AI fingerprints stripped