Company - SYMEUS LTD

England, United Kingdom · Hybrid | Industry - Finance

Operating as a senior individual contributor, leading the design, deployment, and operation of production LLM and RAG systems end-to-end. AI / LLM Systems (Primary): Led development of production-grade LLM and RAG solutions, owning ingestion, embedding pipelines, retrieval, prompt orchestration, evaluation, deployment, and monitoring Built LLM-powered intelligent search, chatbots, and ranking workflows used in internal products Designed and implemented LLM evaluation frameworks (precision/recall, F1, BLEU, ROUGE, relevance & faithfulness metrics) to measure quality and reduce hallucinations Implemented prompt strategies, grounding logic, guardrails, and output validation to improve reliability and safety in live systems Fine-tuned transformer and LLM models on domain-specific data to improve retrieval accuracy and contextual relevance Built scalable data preparation pipelines (cleaning, tokenization, chunking, embeddings) for training and fine-tuning Ran A/B tests and benchmarks across models, embeddings, retrievers, and prompt variants to optimize quality, latency, and cost Machine Learning & Applied Analytics (Secondary): Developed forecasting models for revenue, traffic, and user activity using Python and ML frameworks Built statistical models to support scenario planning and financial analysis Created automated insights pipelines using GA4, GSC, and internal data sources Data Engineering & Platform: Built scalable data pipelines using Apache Spark and Azure Synapse to support near real-time analytics and ML workloads Optimized complex SQL / T-SQL queries (CTEs, window functions, indexing) to improve performance Designed KPI dashboards in Power BI and Streamlit using advanced DAX and Python Improved data quality through governance, validation, and automated checks, reducing reporting turnaround time by ~40%

Company - Alstom

Bengaluru, India · On-site | Industry - Railways

Applied Machine Learning & Predictive Systems: Developed production machine learning models for predictive maintenance, including Remaining Useful Life (RUL) estimation for critical train components Built time-series and survival analysis models to predict failures, degradation, and maintenance needs across multiple subsystems Analyzed high-volume telemetry data to identify failure patterns, sensor drift, and anomalous behavior in operational environments Designed component-level health indicators and engineered features that improved prediction accuracy and model stability Data Engineering & Model Integration Implemented robust data validation, preprocessing, and feature pipelines to ensure reliability of sensor and operational data Integrated predictive models into operational reporting and decision-support systems, enabling faster and more informed maintenance decisions Supported condition-based maintenance strategies that improved fleet availability and reduced unplanned downtime Analytics Platforms & Visualization Built dashboards in Shiny, Qlik Sense, and Tableau to visualize asset health, predictions, and maintenance KPIs Contributed to the setup of the operations center by delivering KPI-driven visualizations and automated model outputs Applied IEC 62541 standards to improve data acquisition consistency and interoperability across systems

Company - Alstom

Bengaluru, India · On-site | Industry - Railways

Built an early semantic search engine using Elasticsearch and Python (Flask), supporting information retrieval use cases Developed automated ETL pipelines using Apache Airflow Contributed to Spark and Kafka streaming workflows for real-time telemetry data processing Built RPA workflows to automate SAP-based maintenance data handling

Company - Pi Revolutions

Bengaluru, India · On-site | Industry - Retail Tech

Supported data analysis and automation workflows for NFC-enabled billing kiosks Performed exploratory analysis to support process improvements

Company - Alstom

Bengaluru, India · On-site | Industry - Railways

Automated routine business processes using VBA in Excel, ensuring compliance with internal data standards and improving efficiency, which saved more than 10 hours of reporting work each week