Sandeep Kshirsagar
Summary
Microsoft, Google and Databricks certified Lead AI/ML Engineer with a specialization in ML and production grade MLOps. Proven track record in architecting end to end ML lifecycles from automated data ingestion to CI/CD driven deployments that reduced model go-live time by 55%. Expert in building scalable Agentic RAG systems and multi-agent workflows using Azure AI Foundry, vertex AI and Databricks Mosaic AI, MLflow tracking and governed Unity Catalog pipelines
Overview
16
16
years of professional experience
1
1
Certification
Work history
National Grid ESO
Lead AI/ML Engineer
London, United Kingdom
08.2022 - 01.2026
- Led the design and hands-on development of forecasting ML pipelines and an enterprise PEF MLOps platform that enabled scalable training and production deployment of energy forecasting products.
- Cut model go-live time by ~55% by implementing reusable Azure ML python SDK v2 pipelines for data preparation, training, inference and environment promotion.
- Automated end-to-end LLM/ML lifecycle (ingestion, training, deployment, monitoring) using Azure DevOps CI/CD with controlled promotion across Dev, Staging and Production environments.
- Implemented a multi-agent AI pipeline to extract and summarize energy forecasts for control room users with timely insights for efficient grid management.
- Developed an Agentic RAG system on Azure AI Foundry using LangGraph, Azure AI Search to perform multi step reasoning and tool calling for weather and energy forecast queries.
British Telecommunications (BT Group)
AI& ML Engineer
Birmingham, United Kingdom
12.2021 - 08.2022
- Developed knowledge Extraction pipeline for large-scale text corpus with Semantic Extraction using pretrained T5 model, BERT word-sense disambiguation(WSD) and Constituent Parsing with AllenNLP.
- Optimized T5 model to reduce model size by 3x and increase the inference speed up-to 5x by converting into ONNX and quantize using FastT5 library.
- Deployed CSIM (customer service insight and management) ML multiple multivariate timeseries model which help identify the expensive customer journeys to Openreach and Proactively intervene them, resulted into £11m cost saving.
TechMahindra
Lead Data Scientist
Birmingham, United Kingdom
06.2019 - 12.2021
- Implemented and deployed a Hot Start Predictor Model to forecast the cooling trajectory for each turbine shutdown, utilizing historical PI data. This led to a notable improvement in accuracy and a significant financial gain.
- Developed energy Consumption Forecasting model using Prophet to segment customers based on their energy consumption patterns and to forecast future energy consumption. The use of Prophet resulted in a 8% improvement in accuracy compared to traditional methods.
- Designed and developed a product recommendation engine which resulted in an 10% increase in click-through rate (CTR).
Virtusa
Senior Data Scientist
Hyderabad, India
09.2018 - 05.2019
- Automated the process of directing the customer issues to the respective development team to increase revenue by 20%.
- Automated ML models deployment using Azure CI/CD pipelines and monitored the data drift in production.
Medtronic
Software Engineer II
Hyderabad, India
07.2015 - 09.2018
- Built ML classifier to identify defective piece of signia handle produced during manufacturing using product images.
- Analyzed the reload transactions data and designed an analytical framework that will help customer to identify segments of customers based on the reload usage and, further classify the customers based on the segments.
Rockwell Collins
Tech Lead
Hyderabad, India
09.2009 - 07.2015
- Developed Hmap multithreaded display application using C++ POSIX library on Green Hills Integrity RTOS .
Education
PG Diploma - AI and ML
Carnegie Mellon University
USABachelor's - Computer Science
SGGSIE&T
IndiaSkills
- Azure Machine Learning (SDK v2)
- Azure AI Foundry
- Vertex AI
- Hugging Face Transformers & Scikit-learn
- Azure OpenAI Service
- LangChain, LangGraph, Semantic Kernel, Prompt Flow
- Model Deployment & Monitoring
- PySpark, Databricks Jobs API & Workflows
- Databrick Asset Bundles
- Time Series Forecasting
- MLflow, TensorFlow, PyTorch, Kubeflow
- Azure DevOps (CI/CD)
Certification
- Microsoft certified Azure Data Scientist
- Google Cloud Certified Professional Data Engineer
References
References available upon request.
Timeline
National Grid ESO
Lead AI/ML Engineer
08.2022 - 01.2026
British Telecommunications (BT Group)
AI& ML Engineer
12.2021 - 08.2022
TechMahindra
Lead Data Scientist
06.2019 - 12.2021
Virtusa
Senior Data Scientist
09.2018 - 05.2019
Medtronic
Software Engineer II
07.2015 - 09.2018
Rockwell Collins
Tech Lead
09.2009 - 07.2015
PG Diploma - AI and ML
Carnegie Mellon University
Bachelor's - Computer Science
SGGSIE&T