AI Case Study 2

ML infrastructure + research-to-prod hires for a real-time emotional AI company

Sector: Multimodal AI avatars

Stage: Seed ($10M, Accel + South Park Commons + Lightspeed)

Location: Seattle, on-site 5 days

Roles: ML Research Engineer + ML Infra Engineer + Operations Lead

The brief

The client is building the first human foundation model — an AI that understands and expresses emotion in real time across speech, facial expression, and body language. Founded by ex-Apple PhDs from MIT, UW, and Oxford. They explicitly asked us for “really strong Video SWE from Robotics, AR/VR, Drones, and Gaming companies” — a target pool most AI recruiters wouldn’t even think to hit.

The challenge

Three concurrent roles at up to $450K base each, all requiring a willingness to relocate to Seattle and work 5 days on-site. The ML infra role had a 2+ YoE floor, meaning we couldn’t just source senior staff engineers — we needed sharp early-career infra engineers with WebRTC or video/audio production experience.

What we did

• Widened the target company set beyond the usual AI-avatar suspects (Synthesia, HeyGen, Character.ai) to include Boston Dynamics, Figure, 1X, Meta Reality Labs, Apple Vision Pro, Niantic, DJI, Skydio, Roblox, Unity, and Epic
• Tapped the team’s internal network: the client’s Founding Engineer is ex-Discord, opening a specific referral path into WebRTC and real-time video engineering communities
• Ran parallel searches for all three roles simultaneously to avoid candidates competing with themselves

Outcome

Searches progressed in parallel across research, infrastructure, and operations, with shortlists covering all three target verticals (robotics, gaming, and consumer video).