Shiv Tripathi

Machine Learning Engineer with 7+ years of software engineering experience, specializing in Machine Learning, Deep Learning, NLP, and Generative AI. Experienced in building and deploying end-to-end ML solutions including LLM fine-tuning, RAG architectures, synthetic data generation, LLM evaluation, Agentic AI, and guardrails using LangChain and LangGraph. Strong hands-on experience with Python, model evaluation, embeddings, vector retrieval, AWS, Docker, Terraform, and CI/CD.

Machine Learning Engineer (MLE) - SERVICENOW - Hyderabad, India

(2023-12)

• Collaborated on Content Moderation & Guardrails for the NowLLM Model across the end-to-end ML lifecycle including LLM fine-tuning, synthetic data generation, model evaluation, threshold optimization, real-time monitoring, and iterative model improvement; analyzed fault patterns in the BERT base model to improve model reliability and moderation performance.
• Designed and evaluated model threshold strategies across multiple versions, conducting POCs to identify optimal operating points and validate precision, recall, false-positive/false-negative trade-offs, and overall model performance.
• Built an automated synthetic-data generation and evaluation pipeline using LangGraph, LLM-as-a-Judge, and multi-model orchestration, generating English and multilingual training and evaluation datasets; automated ~90% of the data-generation workflow, reducing manual effort and accelerating model experimentation and evaluation.
• Designed and deployed a Multi-Agent Tool-Calling system for agentic workflows across HR, ITSM, and CSM use cases; architected agent workflows and orchestration, optimized tool-selection strategies, and defined structured output patterns, resulting in approximately 50% reduction in operational effort.
• Built and contributed to an end-to-end RAG system for Content Moderation and Guardrails using research papers, moderation guidelines, and guardrail rules as the knowledge base; implemented document ingestion, chunking, embedding generation, vector retrieval, and LLM-based generation for guardrail rule definition and context-grounded synthetic data generation. Contributed to Docker-based containerization, Terraform-based infrastructure provisioning, and AWS deployment.

Software Development Engineer in Test (SDET) - EPAM SYSTEMS - Hyderabad, India

(2018-09 - 2023-12)

• Designed and developed a cloud-based AWS automation framework for testing and validating distributed AWS microservices across multiple environments, integrating Lambda, SQS, SNS, SES, S3, DynamoDB, RDS, CloudWatch, and EC2.
• Built and automated cloud infrastructure using Terraform and Docker, enabling consistent provisioning and testing of AWS-based environments; leveraged LocalStack to emulate AWS services and reduce dependency on live AWS infrastructure, reducing cloud testing costs.
• Designed a Docker-based mock environment for the production Trino database to support automated data setup and integration testing, and integrated the solution with AWS ECS (Elastic Container Service) to validate containerized workloads in a cloud environment.
• Engineered CI/CD-ready automation capabilities using Java/Python, JUnit5, Robot Framework, Maven, Jenkins, and GitLab, including parallel execution, isolated logging, reporting, and automated test orchestration to improve reliability and efficiency of cloud-based validation workflows.

Post Graduate Program - AI and ML - International Institute of Information Technology Hyderabad (IIITH) (2024 - 2025)

B.E - CSE - Shri Ramdeobaba College Of Engineering And Management (2014 - 2018)