AI Session-Based Recommendation Model Developer

Talmion Data

Remote · $130k – $160k / year · Full-time · Posted 4 months ago

This role is no longer accepting applications

Don't miss the next role like this

Create a free profile with your skills. We match you to new AI roles as they open, and companies hiring can contact you directly.

Create free profile, 2 minutes

Similar open roles you can apply to now:

See all open AI jobs

This role is no longer accepting applications

Don't miss the next role like this

Create a free profile with your skills. We match you to new AI roles as they open, and companies hiring can contact you directly.

Create free profile, 2 minutes

Similar open roles you can apply to now:

See all open AI jobs

About the role

Session-based recommendation addresses the cold-start limitation of collaborative filtering: when a new user arrives or an existing user starts a fresh session without clear long-term profile signals, what should you recommend? Session-based models work from the sequence of interactions within the current session — the items browsed, the dwell time on each, the search queries, the content abandoned — to infer immediate intent and surface the next most relevant item without needing historical data. Our personalization platform serves 60 media and e-commerce clients with 85 million monthly active users. Session-based recommendation is the capability that matters most for acquisition traffic: visitors arriving from paid search or social media arrive without browsing history, and the recommendation quality for those users directly affects conversion rate. Our current real-time layer relies on collaborative filtering signals that fail for new sessions. You'll lead session-based recommendation model development: designing transformer-based architectures that model within-session interaction sequences, implementing hybrid approaches that combine session signals with available long-term profile context, and building the serving infrastructure that delivers personalized recommendations under 50ms latency constraints for first-time sessions.

Fully remote.

Responsibilities

Design transformer-based session recommendation architectures modeling within-session interaction sequences
Implement hybrid models combining real-time session signals with available long-term profile context
Build session intent inference identifying whether a session is in browse, compare, or purchase mode
Develop cold-start handling for users with zero interaction history before current session
Create serving infrastructure delivering session recommendations under 50ms latency
Monitor session recommendation lift against non-personalized fallbacks as primary quality metric

Requirements

6+ years in recommender systems or ML engineering with session modeling or sequential recommendation experience
Deep understanding of attention-based sequence models and their application to recommendation
Strong Python and PyTorch skills with real-time ML serving at scale
Experience with cold-start recommendation strategies and new user acquisition traffic
Track record demonstrating measurable recommendation quality improvements on cold-start cohorts