Oct 2023 – Nov 2024 · 1 yr 1 mo · Greater London, England, United Kingdom

• Cody -- context engineering; fine-tuning; code completions; auto-complete; code chat
• making coding agents work at enterprise scale
• Improving FIM Code Completions via Context & Curriculum Based Learning
• https://arxiv.org/pdf/2412.16589
• WSDM 2025 paper

Jan 2022 – Sep 2023 · 1 yr 8 mos · Greater London, England, United Kingdom

• Leading an org of 65+ ML engineers, backend engineers, decision scientists and ML scientists that power the entire Sharechat app, serving recommendations to over 180M MAUs, 50M+ creators and 80M+ content per month. My team's AI efforts are focused on multi-objective balancing of user-creator-strategic business goals, spread across 6 key teams in India and Europe.

Oct 2021 – Jan 2022 · 3 mos

• As an area tech lead, I initiated the balancing science effort across the mission (4 squads, >20 engineers, researchers & data scientists), responsible for roadmapping and implementing the machine learning methods that power Spotify sets & playlists accounting for > 15% overall music consumption. Developed ML strategy and roadmap for multi-stakeholder decisioning, including ML modelling, measurement efforts, trade-off decisioning and system design.
• Published 21 research papers during my time at Spotify.
• Filed 9 patents on machine learning and recommendation topics.
• Supervised 8 PhD interns, and 7 MSc thesis students.
• Founding member of TechResearch & London research lab.
• Delivered significant improvement to key long term user satisfaction metrics, alongside other artist and revenue objectives. Led the deployment of various ML projects. Work covered regularly under Two Sided Marketplace theme in Spotify’s quarterly financial reports.
• Q2 2021 shareholder letter:
• https://s22.q4cdn.com/540910603/files/doc_financials/2021/q2/Shareholder-Letter-Q2-2021_FINAL.pdf
• Q1 2021 shareholder letter:
• https://s22.q4cdn.com/540910603/files/doc_financials/2021/q1/Shareholder-Letter-Q1-2021_FINAL.pdf
• Notable publications:
• 1) [TheWebConf 2022] Mostra: A Flexible Balancing Framework to Trade-off User, Artist and Platform Objectives for Music Sequencing; Emanuele Bugliarello, Rishabh Mehrotra, James Kirk, Mounia Lalmas
• 2) [CIKM 2021] Algorithmic Balancing of Familiarity, Similarity, & Discovery in Music Recommendations; Rishabh Mehrotra
• 3) [NeurIPS Why 2021] Disentangling Causal Effects from Sets of Interventions in the Presence of Unobserved Confounders; Olivier Jeunen, Ciarán M Gilligan-Lee, Rishabh Mehrotra, Mounia Lalmas
• 4) [ICML SubSetML 2021] Multi-objective Recommendations via Submodular Objective Diversification & Counterfactual Sequencing; Rishabh Mehrotra

Mar 2021 – Oct 2021 · 7 mos

• As a tech lead, I lead the playlist creation ML efforts from basic research to production, with >15 product launches touching 350+ million users. I lead the balancing science workstream responsible for roadmapping and implementing the machine learning methods that power recommendation products accounting for > 12% overall music consumption on Spotify. Delivered 8%+ improvement to key long term user satisfaction metrics, alongside other artist and business strategic objectives.
• Researching and deploying ML solutions for multi-stakeholder decisioning:
• Multi-stakeholder recommendations
• Sustainable algorithmic decisions
• Sequential recommendations on Radio & Autoplay
• Objective balancing, long term impact, neural modelling, counterfactual decisioning
• Notable publications:
• 1) [WSDM 2021] Shifting Consumption towards Diverse Content on Music Streaming Platforms; C Hansen, R Mehrotra, C Hansen, B Brost, L Maystre, M Lalmas
• 2) [ISMIR 2021] Multi-Task Learning of Graph-based Inductive Representations of Music Content; A Saravanou, F Tomasi, R Mehrotra, M Lalmas
• 3) [CIKM 2020] Query Understanding for Surfacing Under-served Music Content; F Tomasi, R Mehrotra, A Pappu, J Butepage, B Brost, H Galvao, M Lalmas
• 4)[RecSys 2020] Contextual and Sequential User Embeddings for Large-Scale Music Recommendation; C Hansen, C Hansen, L Maystre, R Mehrotra, B Brost, F Tomasi, M Lalmas
• 5) [RecSys 2020] Investigating Listeners’ Responses to Divergent Recommendations; R Mehrotra, C Shah, B Carterette
• 6) [RecSys 2020] Inferring the Causal Impact of New Track Releases on Music Recommendation Platforms through Counterfactual Predictions; R Mehrotra, P Bhattacharya, M Lalmas

Aug 2019 – Feb 2021 · 1 yr 6 mos

• As a research lead, I led ML research for cross-mission Marketplace objective, focusing on multiobjective ML for recommendations on Spotify Radio & Autoplay. I collaborated with 5+ squads to productionize research on multi-objective models, leading the deployment efforts of ML models to 300M+ users ensuring strategic metric wins.
• Major topics:
• Recommendations in multi-stakeholder marketplaces
• Multi-objective and multi-task learning for recommendations
• User & content intelligence for better personalization
• ML focus areas: Neural ranking, Bandits, RL, submodular optimization
• Application areas: set creation, content ranking, objective balancing, user understanding
• Notable publications:
• 1) [KDD 2020] Bandit based Optimization of Multiple Objectives on a Music Streaming Platform
• Rishabh Mehrotra*, Niannan Xue*, Mounia Lalmas (*equal contribution)
• 2) [KDD 2020] Learning with Limited Labels via Momentum Damped & Differentially Weighted Optimization; Rishabh Mehrotra, Ashish Gupta
• 3) [KDD 2020] Counterfactual Evaluation of Slate Recommendations with Sequential Reward Interactions; James McInerney, Brian Brost, Praveen Chandar, Rishabh Mehrotra, Ben Carterette
• 4) [WWW 2020] Algorithmic Effects on the Diversity of Consumption on Spotify; Ashton Anderson, Lucas Maystre, Ian Anderson, Rishabh Mehrotra, Mounia Lalmas
• 5) [NLDL 2020] Joint Attention Neural Model for Demand Prediction in Online Marketplaces; A Gupta, R Mehrotra

Nov 2017 – Aug 2019 · 1 yr 9 mos

• As a scientist, I have worked on a wide variety of projects on personalization and recommendation topics, including bandit reward modelling, user intent understanding, metrics & evaluation. We developed models for bandit based ranking, understanding user intents, mixed-methods for metric development for user engagement modeling on Spotify homepage. Notable publications from our work include:
• Notable publications:
• 1) [CIKM 2018] Towards a Fair Marketplace: Counterfactual Evaluation of the trade-off between Relevance, Fairness & Satisfaction in Recommendation Systems
• 2) [WWW 2019] Jointly Leveraging Intent and Interaction Signals to Predict User Satisfaction with Slate Recommendations
• 3) [WWW 2019] Deriving User- and Content-specific Rewards for Contextual Bandits; Paolo Dragone, Rishabh Mehrotra, Mounia Lalmas
• 4) [RecSys 2018] Explore, Exploit, and Explain: Personalizing Explainable Recommendations with Bandits; James McInerney, Benjamin Lacker, Samantha Hansen, Karl Higley, Hugues Bouchard, Alois Gruson, Rishabh Mehrotra
• 5) [WWW 2020] Leveraging Behavioral Heterogeneity Across Markets for Cross-Market Training of Recommender Systems; Kevin Roitero, Ben Carterrete, Rishabh Mehrotra, Mounia Lalmas

Jan 2017 – Aug 2017 · 7 mos · London Area, United Kingdom

• Deep technology solutions around understanding user tasks based on cutting-edge machine learning methods which helps businesses to provide users with personalized proactive task assistance.

Jan 2016 – Jan 2016 · 0 mo · Greater New York City Area

• Fairness in Search & Learning algorithms + Deep Sequential Models for Task Satisfaction
• Host: Fernando Diaz, Ahmed Hassan Awadallah

Jul 2015 – Sep 2015 · 2 mos

• Characterizing Cross-Domain Search Behavior

Apr 2015 – Jun 2015 · 2 mos

• Counterfactual estimation of related searches metrics

Feb 2014 – Apr 2017 · 3 yrs 2 mos · London, United Kingdom

• My PhD research focussed on developing efficient bayesian & deep models for user understanding, knowledge discovery and decision optimization. I worked on developing task understanding systems for web search and conversational intelligence.

Jun 2013 – Dec 2013 · 6 mos

Jan 2013 – Jun 2013 · 5 mos · Bangalore

• Adaptive learning & adaptive testing; Behavioural modelling in organizational networks

Sep 2012 – Dec 2012 · 3 mos · Greater Houston

• Developed online algorithms for Semi-Coupled Dictionary Learning which incorporates a discriminative objective useful in multi-view learning settings.

May 2012 – Aug 2012 · 3 mos · Canberra, Australian Capital Territory, Australia

• Worked on improving Latent Dirichlet Allocation topic models for microblogs via automatic tweet labeling and pooling.

Jun 2011 – Jun 2011 · 0 mo · Singapore

May 2010 – Jul 2010 · 2 mos

• Developed algorithms for resolving polysemy and synonymy for high-quality information extraction from news articles. Published results at 49th Association of Computational Linguistics ACL HLT Workshop WASSA, Portland, US.