Shiv Saini
Scholar

Shiv Saini

Google Scholar ID: gWocpdkAAAAJ
Adobe Research
Efficient AI-driven systemsCausalityTime Series
Citations & Impact
All-time
Citations
669
 
H-index
14
 
i10-index
22
 
Publications
20
 
Co-authors
19
list available
Publications
20 items
Browse publications on Google Scholar (top-right) ↗
Resume (English only)
Academic Achievements
  • Cache-Craft: Managing Chunk-Caches for Efficient Retrieval-Augmented Generation (SIGMOD, 2025)
  • ReCON: Training-Free Acceleration for Text-to-Image Synthesis with Retrieval of Concept Prompt Trajectories (ECCV, 2024)
  • Approximate Caching for Efficiently Serving Diffusion Models (NSDI, 2024)
  • Outage-Watch: Early Prediction of Outages using Extreme Event Regularizer (ESEC/FSE, 2023)
  • ESRO: Experience Assisted Service Reliability against Outages (ASE, 2023)
  • CausIL: Causal Graph for Instance Level Microservice Data (WWW, 2023)
  • Towards Optimizing Storage Costs on the Cloud (ICDE, 2023)
  • ViSRE: A Unified Visual Analysis Dashboard for Proactive Cloud Outage Management (VISSOFT, 2022)
  • Modeling Causal Impact of Textual Style on a Targeted Goal (WebConf - Poster, 2020)
  • Time Series Forecasting for Cold-Start Items by Learning from Related Items using Memory Networks (WebConf - Poster, 2020)
  • Multiple Treatment Effect Estimation using Deep Generative Model with Task Embedding (The World Wide Web Conference, 2019)
  • Modeling Hint-Taking Behavior and Knowledge State of Students with Multi-Task Learning (International Conference on Educational Data Mining, 2018)
  • Sparse Decomposition for Time Series Forecasting and Anomaly Detection (SDM, 2018)
  • A Non-parametric Approach to the Multi-channel Attribution Problem (WISE 2015)
Research Experience
  • Principal Research Scientist, Bangalore, Adobe Research
Background
  • Research Interests: Time series modeling and causal inference in observational data; Applications: Modeling user behavior, marketing attribution root cause analysis, anomaly detection; Recent research focus: Developing techniques to improve reliability of microservices by forecasting future outages and reducing time to detect root cause of ongoing outages.