We are hiring an AI Engineer to be the lead technical contributor on a personalization and
ranking engagement for a large-scale consumer marketplace. You will set the technical
direction, make the key modeling decisions, and stay hands-on throughout. You will be a senior
technical point of contact with the customer — explaining trade-offs, managing expectations,
and turning results into clear recommendations. You will lead a rigorous, POC-first program:
engineering user-level features from behavioral data, integrating LLM-generated user profiles
into a deep-learning ranking model, and driving the work from offline validation through
production-readiness.
What You’ll Do
• Own the technical strategy for a personalization program on a production
recommendation/ranking system, making the architecture and modeling decisions and
being accountable for the results.
• Stay hands-on: build the features, train the models, run the experiments, and write the
critical code.
• Set the technical bar and support other engineers through design reviews, mentorship,
and pairing.
• Act as a senior technical point of contact with the customer, communicating progress,
risks, and results to both engineers and senior stakeholders, and managing expectations
through ambiguity.
• Design and run a structured, parallel-track proof-of-concept that measures the incremental
lift of GenAI-based profiles over well-engineered behavioral ML features.
• Engineer user-level features from large-scale behavioral data (category/product affinity,
time-of-day and price-sensitivity patterns, per-user click/conversion history, recency-
frequency signals).
• Integrate LLM-generated user profiles into ranking models, including embedding
generation, projection-layer tuning, gating, and ablation to ensure the signal is properly
weighted.
• Own the deep-learning ranking model (multi-task CTR/CVR architectures such as shared-
bottom MTL), including feature integration, hyperparameter optimization (Bayesian/grid
search), and bias correction (position/popularity).
• Define and run the offline evaluation framework — NDCG, MRR, Precision/Recall at K —
with segment-level analysis and ablation studies across user cohorts.
• Establish the path to production: model serving and scheduled inference integration,
shadow-mode testing, A/B framework readiness, and guardrail metrics.
• Deliver clear technical documentation and lead knowledge-transfer sessions so the
customer’s teams can operate and iterate independently after handoff.
Required Qualifications• 10+ years in applied machine learning / data science, with deep hands-on experience in
recommender systems, learning-to-rank, or large-scale personalization.
• Practical experience building with LLMs in production: generating and integrating model-
derived features or profiles, working with embeddings, and reasoning about evaluation,
latency, and cost.
• Experience with Amazon Bedrock or comparable managed LLM platforms for production
inference.
• Hands-on experience with segment- or cohort-based personalization, including measuring
performance at the segment level rather than relying on aggregate metrics.
• Experience designing cold-start strategies for users or items with limited history.
• Strong communication skills — able to explain modeling decisions, trade-offs, and results
clearly to engineers, data scientists, and senior business stakeholders, and to manage
expectations through ambiguity.
• Customer-facing or stakeholder-facing experience: building trust, navigating competing
priorities, and serving as a senior technical voice in high-stakes conversations.
• A track record of technical leadership through mentoring engineers, driving design
decisions, and setting standards.
• Strong track record taking ML models from experimentation to production, owning the
offline-to-online validation story (ranking metrics, ablations, segment analysis, shadow
testing, A/B readiness).
• Deep, hands-on expertise in deep learning for ranking/recommendation — multi-task
learning, embedding-based architectures — with a major framework (TensorFlow or
PyTorch).
• Strong feature engineering on large behavioral datasets using the modern data stack
(PySpark, SQL, distributed data lakes).
• Rigorous experimental methodology — hyperparameter optimization, bias correction, and
a disciplined, hypothesis-driven approach to measuring true lift.
• Hands-on AWS experience across the ML lifecycle, and strong proficiency in Python.
Preferred Qualifications
• Experience personalizing ranking for marketplaces or consumer platforms at scale (e-
commerce, food delivery, media, or similar).
• MLOps maturity: model versioning, monitoring, and reproducible training pipelines.
• Advanced degree in Computer Science, Machine Learning, Statistics, or a related
quantitative field.
• Prior experience in a client-facing consulting or professional-services delivery
environment.
