Aman Jain

Developed an Agent builder using Chain-of-thought (CoT) prompting for agent configuration and tool calls with DSPy framework, creating a feedback-driven chat platform. Implemented guardrails for natural language constraint definition and custom guardrail deployment to the LLM studio. Built an agent evaluation framework using LLM-as-a-judge to test configurations and provide feedback.

Led the development of a search service for company, department, and employee details, optimizing OpenSearch schema to reduce costs by 30% and achieve sub-200ms search latency. Created scalable software architecture for gRPC services, set up a gRPC gateway, and deployed the solution on EKS. Architected data pipelines using AWS Batch and Airflow, enabling seamless implementation of data scrapers, L

Designed and developed platform services for training and inference of large-scale data pipelines using Pachyderm, GitHub CI, and Terraform. Led a team of 7 engineers to create a dataset and model management SDK in Python and PostgreSQL, leveraging AWS S3 for ML lifecycle management. Developed an ML model evaluation framework that enabled rapid testing and deployment for maps pipelines. Engineered

Led the development of large-scale microservices in Golang for search features like Dynamic faceting, query autocomplete, and query suggestion. Scaled search microservices from 2,500 to 25,000 RPS, achieving a 30% cost reduction in Elasticsearch by implementing an auto-scaling Kubernetes operator. Migrated Elasticsearch, data pipeline, and monitoring system to a multi-tenant Kubernetes cluster, au

Grade: 8.2/10

Activities and societies: Head of Development at Entrepreneurship club; Winner, GE Digital IIOT Hackathon and Most Impactful Idea, Smart City Hackathon.

Achieved a CGPA of 8.2/10. Served as the Head of Development at the Entrepreneurship club. Won the GE Digital IIOT Hackathon and received the Most Impactful Idea award at the Smart City Hackathon.

Activities and societies: Predicting bike-sharing pattern (Neural network), Dog breed classifier (CNN), TV script generation(RNN), AWS Sagemaker deployment

Completed projects including predicting bike-sharing patterns using neural networks, developing a dog breed classifier with CNN, and generating TV scripts using RNN. Gained experience with AWS Sagemaker deployment.