Srividya Aluru

MSc Computing (AI & ML) student at Imperial College London, researching adversarial robustness of large language models. Have 4+ years of prior experience as a Data Scientist at Ernst & Young, building production-grade LLM systems, multi-agent architectures, and scalable NLP pipelines, providing a strong engineering foundation for executing research at scale. Eligible for the Graduate visa, with full right to work in the UK.

Data Scientist (Senior Consultant) at Ernst & Young (2021-06 – 2025-07)

• Designed an LLM-powered chatbot that generated on-demand data insights for business stakeholders, reducing cross-team data retrieval time from 2+ hours to under 1 minute across 5 enterprise clients.
• Powered the backend using a multi-agent framework with a Chain-of-Thought (CoT) supervisor to dynamically route complex, multi-step queries across 30+ fragmented tables and generate 140 KPIs.
• Curated custom LLM tools for entity detection, KPI schema lookups, and automated SQL/Python to modularize the operations and built a FAISS-backed RAG pipeline using few-shot prompting reducing hallucinations.
• Achieved 95% accuracy on a custom 1,000+ query benchmark and improved performance through a closed user feedback loop, and shipped production features including conversational memory, and user management.
• Separately built Dashboard GENie, a semantic retrieval system (ChromaDB, LangChain) enabling natural-language navigation across 100+ dashboards, achieving 90%+ retrieval accuracy and reducing manual search time by 75%.
• Estimated annual revenue gain of 5 million (2% of total) by devising strategy for identifying new outlets
• Achieved an 80%+ accuracy in forecasting for high-performing markets by leveraging ML based algorithms and spatial analytics to gain real-time insights into customer purchase locations and market coverage
• Applied data-driven insights to identify and advocate for 5-7 new retail store locations in each strategic market
• Conducted a comprehensive analysis of demographic concentrations for competitors' exclusive brands and multi-brand outlets, contributing valuable insights to the brand's market positioning and identifying focus markets
• Designed and developed ML based sales demand forecasting tool for long-term and short-term horizons
• Improved accuracy by 15% (overall accuracy 90% vs 75%) by implementing a best fit model selection algorithm
• reduced run time by 70% (from 6 to 2 hrs.) by streamlining and optimising data wrangling & ETL processes
• Maintained continuous collaboration with stakeholders for alignment on objective & feedback throughout project

M.Sc. in Computing (Artificial Intelligence & Machine Learning) – Imperial College London (2025-10)

Dual Degree in M.Sc. Economics & B.E. Computer Science – BITS Pilani Hyderabad Campus (2016-08 – 2021-05)