Manisha Singh Email & Phone Number

Toronto, On, Ca

Gta

Integrating LLM capabilities in SOTI productsLed the development of an advanced self-service tool for a Mobile Device Management platform, leveraging both proprietaryand open-source LLMs. Designed a comprehensive architecture to integrate three key components: JavaScript codegeneration(AWS Bedrock/AWS Q), a search module, and a technical information retrieval system. Implemented aretrieval-augmented generation (RAG) methodology using data chunking, Redis vector embeddings, and semantic search toensure highly relevant responses. Fine-tuned models with QLora and deployed them with vllm on EKS to enhance performanceand scalability. Chatbot delivered conversational assistance for technical inquiries, device profile analytics, and productfunctionality issues, significantly improving operational efficiency and user experience.Customer Support Improvement● Utilized Large Language Models (LLM) to identify and address urgent technical support cases.● Leveraged GPT to identify and prioritize urgent technical support cases based on sentiment analysis and contextualunderstanding, improving response times by 25% and boosting customer satisfaction scores.● Developed LLM-powered systems for case summarization and QA automation reducing manual effort by 40%, boostingagent productivity and consistent customer experience and ensuring consistent customer experience across channels.● Enabled estimated monthly cost savings of $55,000 through reduced resolution times.● Integrated Salesforce, and JIRA with ChatGPT utilizing software engineering best practices, API development togenerate insights from unstructured data, eliminating the need for extensive manual reporting and enhancing supportteam efficiency.HR Talent ToolLed various projects using NLP and GPT-driven HR Talent Tool to streamline People and Culture initiatives. Simplified daily HRtasks including Resume shortlisting, custom LinkedIn messages optimizing processes and increasing operational efficiency.

Greater Toronto Area, Canada

Cost-Saving Forecasting SolutionDeveloped and deployed machine learning models for add-on recommendations and renewal predictions, leveraging cloudserver activity and customer interactions. Implemented a robust MLOps pipeline using MLflow for experiment tracking,model registry, and deployment. This included logging parameters, metrics, and artifacts, versioning models, and deployingto production environments. These efforts contributed to a $1 million increase in add-ons profit while optimizing resourceallocation.● Server Performance & Utilization EnhancementProactively identified and resolved CPU, RAM, and storage bottlenecks on client servers using 90-day performance data,optimizing resource utilization and enabling cost-effective downsizing/shutdown recommendations.

Greater Toronto Area, Canada

->Developed, designed, and deployed a machine learning solution to forecast add-ons and renewal of device licenses based on cloud server activity and customer interactions. The company’s add-ons profit increased by $1 million.->Worked on feature engineering of various features describing time series characteristics, feature selection using classical filter-based statistical techniques, and model-based selection.->Used classical classification algorithms as well as tree-based ensembles on various datasets and combined them using different approaches (including Bayesian approach, modified maximum voting classifier, and non-correlated averaging)->Build ETL pipelines for feature engineering and scheduling model predictions.

Toronto, Ontario, Canada

Device Management Solution->Worked on finding the quantitative relationship between discharge rate and its KPIs, using statistical modeling and machine learning techniques->Utilized various machine learning methodologies for data smoothing and time series forecasting to formulate discharge rate per application->Implemented time series forecasting techniques for battery life prediction based on real-time usage traces.

Toronto, Canada Area

CSC343 - Introduction to Databases (Fall 2019 and Winter 2020)

Gurgaon, India

->Worked in Business Intelligence domain from ETL to Data Warehousing Architecture and Design and Analytics development->Extensively worked on Data Science Studio with a deep understanding of data modeling->Performing a series of quality inspections, alterations and suggested enhancements based on different User Acceptance Testing cycles->Preparing Test Case, Best Practices and Standard Operation Procedure documents with a varying range of cases and instructions, covering both the front end implementation and the backend architecture, for error-free delivery and smooth operation cycles.->Developing, Optimizing and automating the operations management process using Python->Regular calls for interaction with the onshore team for design and development review, feedback gathering and daily status intimation

Gurgaon, India

Collaborated with BI leadership of a US-based healthcare firm to deliver analysis-ready data repositories, integrating multi-vendor sales and digital marketing data, implementing various algorithms for data standardization. This reduced the analysis time from hours to minutes and also enabled data democratization

Dehra Dun Area, India

Explored advanced android concepts and developed an application for the employees of India’s leading PSU, that allows anyone to search for available bus transport options within the city, along with their route and timings

Mscac, Computer Science

Courses taken: 1. Machine Learning 2. Machine Learning for Health 3. Topics in Multidisciplinary HCI: Designing Intelligent Self-Improving.

Bachelor Of Engineering - Be, Computer Science, 9.27

Activities and Societies: • Event Head at Goonj’17: Head of Byte the Hints Coding competition at techno-cultural fest. • Team Leader.