Rushikesh Joshi

Software Engineer at SAP Concur

Building multi-agent AI architectures and ML recommendation systems. Currently working on agent orchestration for travel planning and recommendation engines that serve millions of users. Go, Java, distributed systems.

Rushikesh Joshi - Software Engineer
Scroll to explore

Background

7 years building backend systems. Currently focused on multi-agent AI — designing architectures where multiple specialized agents coordinate to solve complex tasks like end-to-end travel planning.

At SAP Concur, I've built recommendation engines for hotels, flights, and rooms, plus agentic systems for parsing requests, auto-completing search criteria, and generating itineraries. The interesting part is making these agents work reliably at scale. Before this, I built transpilers at SAP Labs that decoupled our analytics platform from HANA.

7
Years
in production systems
60%
Accuracy
on hotel recommendations
5x
Faster
SAC uninstallation after rewrite (2x installation)

Professional Experience

Senior Software Engineer - II

SAP Concur

Mar 2023 - Present

  • Designed and implemented a robust multi-agent, fault-tolerant, durable, and scalable Agentic AI architecture for planning, shopping, and booking business travel
  • Enhanced business traveler experience by reducing time-on-task using Machine Learning to develop personalized:
    • - Hotel recommendation engine with ~60% accuracy
    • - Hotel room recommendation engine with ~40% accuracy, powered by LLMs to transform unstructured room descriptions into structured room categories
    • - Flight recommendation engine with ~50% accuracy
    • - Historical data to predict trip origin location and hotel and rental-car needs per location
  • Improved PSAT score by 5% and lowered regressions to 0 by engineering:
    • - A/B testing framework and a shadow (background) deployment framework enabling testing multiple ML models in parallel in production systems
    • - Event-driven real-time monitoring system with data visualization dashboards to track accuracies of multiple ML models
    • - AI-powered GitHub workflow to create pull request summaries and code reviews
    • - AI-powered GitHub workflow to classify pull requests by complexity and risk, with the ability to auto-approve simple, low-risk pull requests
Go Genkit LangGraph GraphQL Redis/ValKey AWS RDS (Postgres) AWS DynamoDB AWS SNS AWS SQS Dynatrace Kubernetes

Software Engineer - III

SAP Labs

Feb 2021 - Mar 2023

  • Re-engineered SAP Analytics Cloud by rewriting XSJS to modular TypeScript, improving installation speeds by 200% and uninstallation by 500%
  • Decoupled SAC from HANA and enabled SQL installations by developing transpilers:
    • - HANA database artifacts to JSON object schema definitions
    • - JSON object schemas to SQL queries
TypeScript NodeJS Redis SQL Kubernetes

Software Engineer I

Susquehanna International Group

Sept 2020 - Feb 2021

  • Decreased testing costs by 50% by developing a stock exchange simulator for traders to test algorithms
C#

Software Engineer I

SAP Labs

Sept 2017 - May 2019

  • Increased customer satisfaction and accelerated time-to-market by decomposing monolithic banking platform into RESTful microservices
Java Spring Hazelcast SQL Maven Gradle

Skills & Technologies

Languages

Cloud & Databases

DevOps & Tools

AI & Frameworks

Double-click or press Enter on any skill for a fun fact!

Education

Masters in Computer Sciences - Data Sciences

Trinity College Dublin (University of Dublin)

First Class with Distinction

Sept 2019 - Aug 2020 • Dublin, Ireland

Bachelor of Engineering - Computer Engineering

Maharashtra Institute of Technology (University of Pune)

First Class with Distinction

Jun 2013 - May 2017 • Pune, India

Get In Touch

If you're working on multi-agent systems, ML infrastructure, or recommendation engines — happy to chat. Always interested in discussing agent architectures and how to make them reliable in production.

Send me a message