AJAY SUMANTH

Mountain View

Summary

Driven problem-solver with passion for designing innovative software solutions. Possesses solid understanding of programming languages and development frameworks. Committed to delivering impactful software products that enhance user experiences and drive business success.

Overview

years of professional experience

Work History

Software Developer

SAP

Palo Alto, CA

01.2025 - Current

Added web search and file upload capabilities to Joule, SAP’s AI copilot
Scaled the Joule Message Generation Module to handle of large values in response generation by implementing semantic placeholder-based compression techniques
Contributed to the Joule Compiler project, developing the CRM Digital Assistant and creating a new Procurement Assistant capability to extend Joule’s domain coverage across SAP lines of business
Researched decoder-based critic models for reasoning over tree-structured problems, exploring how token-level success probabilities can guide efficient branch pruning without full exploration
Developed the core evaluation engine for a generative AI copilot that converts natural language business scenarios into schema.cds and services.cds files, enabling automated schema and service generation in the SAP Cloud Application Programming model
Built a Jenkins integration to automate the end-to-end Playwright test pipeline, generating an HTML test dashboard to visualize pipeline execution and results
Built a copilot that converts user-provided design instructions into fully structured SAP Build Workzone pages, enabling rapid, code-free UI creation for internal business portals
Trained and fine-tuned model to generate JavaScript handlers for CAP-based services, enabling end-to-end automation of backend logic from natural language inputs

Software Developer Intern

SAP Ariba

Palo Alto, CA

06.2024 - 12.2024

Designed and implemented multiple schemas to support regional and business unit-specific supplier management workflows in SAP Ariba
Used BERT embeddings with a support vector classifier (SVC) to generate similarity scores, matching new ServiceNow tickets with previous ones for efficient developer assignment
Integrated with the ServiceNow API to fetch, assign, and update tickets based on similarity analysis, enabling real-time workflow automation
Fixed critical security vulnerability in multi-tenant application by implementing proper thread-local context clearing, preventing unauthorized cross-realm data access in distributed systems

Software Engineer Intern

Country Financial

Bloomington, IL

05.2023 - 12.2023

Developed a machine learning model using XGBoost to determine fraud in auto insurance claims, significantly increasing the accuracy of insurance fraud detection with a potential to save millions of dollars
Utilized Delphix platform to streamline data provisioning for development, testing, and analytics, resulting in a 30% reduction in data duplication and storage costs
Designed and optimized database schemas using DB2 and Oracle, improving query performance by 20%
Developed scripts in Linux using shell scripting to automate database backup and restore operations

Education

Master’s - Computer Science

Rice University

Houston, TX

12.2026

Bachelor’s - Computer Science & Engineering

University of Michigan

Ann Arbor, MI

12.2024

Skills

Languages: C/C, Python, Java, JavaScript, Julia, MATLAB, C#, Assembly, SQL

Computer: Apache Kafka, Apache Spark, AWS, Docker, FastAPI, Git (Version Control), Jenkins, Kubernetes, Linux, NumPy, Pandas, PyTorch, Retrieval-Augmented Generation (RAG), Scikit-learn, Tableau, TensorFlow, XGBoost

Websites

Timeline