Research
Research on the systems around the model.
Deep experience in systems intelligence, autonomous agents, causal reasoning, reinforcement learning, large-scale personalization, and marketplace optimization.
NeurIPS · ICLR · MICCAI · KDD · WWW · SIGIR · WSDM · RecSys · CIKM
Systems Intelligence & Autonomous Agents
Lifelong learning, multi-agent coordination, agent safety, and system ownership, the research foundations for compounding enterprise AI.
- System Ownership as a Lifelong Learning Problem: A Formulation for Long-Horizon Software MaintenanceP. Trochim, S. Pan, V. ChandelaPre-printStealth
- Informational Individuality and Behavioural Consistency: A Theory Framework of Lifelong Learning for LLM AgentsS. Pan, P. Trochim, V. Chandela, R. MehrotraPre-printStealth
- How Task Structure Limits Multi-Agent Success: An Information-Theoretic AnalysisS. Pan, M. LuoOpenReview 2026
- Programmatic Process Rewards Improve the Reliability of Agent-Safety Reinforcement LearningS. Pan, R. MehrotraPre-printStealth
- Mini-uber: Hold-Probe Evaluation for Multi-Regime Agent TasksS. Pan, R. MehrotraPre-printStealth
- Spectrum-Anchored Updates Preserve Plasticity in Continual LearningS. Pan, X. GuanPre-printStealth
- AutoGT: Distilling High Quality Evaluation Ground Truth from Heterogeneous Sources on Knowledge-Intensive TasksS. Pan, S. Saket, R. MehrotraPre-printStealth
- RCA Playbooks: Probabilistic Reconstruction of Diagnostic Workflows from SQL Query GraphsS. Saket, S. Dhar, R. MehrotraPre-printStealth
- Prescriptive Cheatsheets: Structured Artifacts via Evidence-Aware Submodular SynthesisS. Saket, S. Dhar, R. MehrotraPre-printStealth
- Semantic Factorization of Analytical SQL WorkloadsS. Saket, S. Dhar, R. MehrotraPre-printStealth
AI-Powered Engineering & Code Intelligence
Intelligent systems that understand codebases, retrieve context, and assist engineers with production-grade code recommendations.
Intelligent Decision Systems & Reinforcement Learning
Causal reasoning, model-based planning, and bandit optimization, algorithmic foundations for systems that learn from decisions and compound over time.
- Disentangling Causal Effects from Sets of Interventions in the Presence of Unobserved ConfoundersO. Jeunen, C. Gilligan-Lee, R. Mehrotra, M. LalmasNeurIPS 2022
- Evaluating Model-Based Planning and Planner Amortization for Continuous ControlA. Byravan, L. Hasenclever, P. Trochim, M. Mirza, A. Ialongo, Y. Tassa, J. Springenberg, A. Abdolmaleki, N. Heess, J. Merel, M. RiedmillerICLR 2022
- Teaching Pathology Foundation Models to Accurately Predict Gene Expression with Parameter Efficient Knowledge TransferS. Pan, J. Chen, M. SecrierMICCAI 2025
- Bandit based Optimization of Multiple Objectives on a Music Streaming PlatformR. Mehrotra, N. Xue, M. LalmasKDD 2020
- Counterfactual Evaluation of Slate Recommendations with Sequential Reward InteractionsJ. McInerney, B. Brost, P. Chandar, R. Mehrotra, B. CarteretteKDD 2020
- Ad-load Balancing via Off-policy Learning in a Content MarketplaceH. Sagtani, M. Jhawar, R. Mehrotra, O. JeunenWSDM 2024
- Deriving User- and Content-specific Rewards for Contextual BanditsP. Dragone, R. Mehrotra, M. LalmasWWW 2019
- Explore, Exploit, and Explain: Personalizing Explainable Recommendations with BanditsJ. McInerney, B. Lacker, S. Hansen, K. Higley, H. Bouchard, A. Gruson, R. MehrotraRecSys 2018
- Semi-analytical Industrial Cooling System Model for Reinforcement LearningY. Chervonyi, P. Dutta, P. Trochim, O. Voicu, C. Paduraru, C. Qian, E. Karagozler, J. Davis, R. Chippendale, G. Bajaj, S. Witherspoon, J. LuoarXiv 2022 · DeepMind
Large-Scale Personalization & Recommendation
Architecting ML systems at 100M+ user scale, embeddings, ranking, sequencing, and real-time serving across Spotify, ShareChat, and Seekho.
- Dimension Mask Layer: Optimizing Embedding Efficiency for Scalable ID-based ModelsS. Saket, I. Ihara, V. Sharma, D. KalimWWW 2025
- Monitoring the Evolution of Behavioural Embeddings in Social Media RecommendationS. Saket, O. Jeunen, D. KalimSIGIR 2024
- On Gradient Boosted Decision Trees and Neural Rankers: Short-Video Recommendations at ShareChatO. Jeunen, H. Sagtani, H. Doi, R. Karimov, N. Pokharna, D. Kalim, A. Ustimenko, C. Green, R. Mehrotra, W. ShiFIRE 2023
- MEMER, Multimodal Encoder for Multi-signal Early-stage RecommendationsM. Agarwal, S. Saket, R. MehrotraWWW 2023
- Formulating Video Watch Success Signals for Recommendations on Short Video PlatformsS. Saket, S. Velugoti, R. MehrotraLERI@RecSys 2023
- Exploiting Sequential Music Preferences via Optimisation-Based SequencingD. Moor, Y. Yuan, R. Mehrotra, Z. Dai, M. LalmasCIKM 2023
- Contextual and Sequential User Embeddings for Large-Scale Music RecommendationC. Hansen, C. Hansen, L. Maystre, R. Mehrotra, B. Brost, F. Tomasi, M. LalmasRecSys 2020
- Algorithmic Balancing of Familiarity, Similarity, & Discovery in Music RecommendationsR. MehrotraCIKM 2021
- Shifting Consumption towards Diverse Content on Music Streaming PlatformsC. Hansen, R. Mehrotra, C. Hansen, B. Brost, L. Maystre, M. LalmasWSDM 2021
- Crafting Tomorrow: The Influence of Design Choices on Fresh Content in Social Media RecommendationS. Saket, M. Agarwal, R. MehrotraarXiv 2024
Multi-Stakeholder Optimization & Marketplace Intelligence
Balancing competing objectives across users, creators, and platforms, fairness, multi-sided value, and system-level trade-offs.
- Mostra: A Flexible Balancing Framework to Trade-off User, Artist and Platform Objectives for Music SequencingE. Bugliarello, R. Mehrotra, J. Kirk, M. LalmasWWW 2022
- Towards a Fair Marketplace: Counterfactual Evaluation of the Trade-off between Relevance, Fairness & SatisfactionR. Mehrotra, J. McInerney, H. Bouchard, M. Lalmas, F. DiazCIKM 2018
- Jointly Leveraging Intent and Interaction Signals to Predict User Satisfaction with Slate RecommendationsR. Mehrotra, M. Lalmas, D. Kenney, T. Lim-Meng, G. HashemianWWW 2019
- The Multisided Complexity of Fairness in Recommender SystemsN. Sonboli, R. Burke, M. Ekstrand, R. MehrotraAI Magazine 2022
- Quantifying and Leveraging User Fatigue for Interventions in Recommender SystemsH. Sagtani, M. Jhawar, A. Gupta, R. MehrotraSIGIR 2023
- Inferring the Causal Impact of New Track Releases on Music Recommendation Platforms through Counterfactual PredictionsR. Mehrotra, P. Bhattacharya, M. LalmasRecSys 2020
- Algorithmic Effects on the Diversity of Consumption on SpotifyA. Anderson, L. Maystre, I. Anderson, R. Mehrotra, M. LalmasWWW 2020
- Learning with Limited Labels via Momentum Damped & Differentially Weighted OptimizationR. Mehrotra, A. GuptaKDD 2020
Papers marked Stealth are pre-prints not yet public.
