Varsha Joshi

Sr. Software Engineer at Amazon

Santa Cruz, California, United States

About

Senior Software Engineer with 8+ years of experience in backend systems, distributed services, and applied machine learning. Proven track record in delivering scalable solutions and developing generative AI applications to optimize workflows and automate complex decision systems. Skilled in Java, Python, AWS, and modern ML frameworks with strong focus on performance, security, and compliance automation.

Experience

  • Amazon (Full-time · 5 yrs 9 mos)
    • Senior Software Engineer
      Jul 2023 - Present · 3 yrs

      - Driving large-scale compliance automation and ML initiatives to enhance efficiency and scalability in regulatory document processing. - LLM-powered Classification Insights: Designed and deployed a system that leverages Large Language Models (LLMs) to generate classification reasons for reviewers, improving handling time by 12% and enabling measurable headcount savings for customers. - Policy Translation Agent: Built an LLM-based policy generation agent that converts customer regulatory documents into structured system policies using AWS Bedrock, knowledge bases, and tool orchestration. Achieved 95% precision and 80% coverage, reducing regulation onboarding time from 15 days to under 1 day. - Partner with product and leadership teams to define the long-term ML roadmap and shape the vision for compliance automation.

    • Software Development Engineer
      Oct 2020 - Jun 2023 · 2 yrs 9 mos

      Delivered scalable backend and ML solutions across Amazon’s Identity and Compliance platforms to strengthen data security and reduce manual review overhead. Anomaly Detection: - Built systems to detect and auto-correct anomalies in eCommerce data using SQS, Beanstalk, and DynamoDB, reducing manual review load by 9%. - Applied Spark and Scala for ML-based review backlog reduction and proactive anomaly detection to optimize reviewer bandwidth.

  • Member of Technical Staff at VMware
    Apr 2018 - Oct 2020 · 2 yrs 7 mos

    - Worked on building an end to end pipeline to efficiently get metadata of processes running on OS. Implemented a hashing mechanism to uniquely identify each behavior to be mapped for analysis on the cloud. The high traffic data is collected and analysed to draw time based recommendations for application security. - Worked on event classification using similarity reduction for the networking traffic. Implemented arithmetic averaging to determine severity of the incoming alerts reducing false positives by thirteen percent

  • Graduate Teaching Assistant at University of California, Santa Cruz
    Sep 2017 - Mar 2018 · 7 mos

    CMPS 111 - Introduction to Operating Systems Teaching assistant for the class of 150 students. Conducted labs, sessions and helped students with OS concepts including scheduling, synchronization, deadlocks and file systems on OS.

  • Sr. Sw. Developer Intern at Visa
    Jun 2017 - Sep 2017 · 4 mos

    Worked as a summer intern in Risk Systems Analysis Team at Visa. The team maintains fraud detection systems. Designed a graphical webapp for internal dashboard making data easily accessible for system analysis Improved cache implementation to handle orphan data resulting in reliable data handling

  • Software Developer at Barclays Investment Bank
    Aug 2014 - Jul 2016 · 2 yrs

    Graduate Development Program 2014-2015 Software Developer – Algorithmic and Electronic Trading Dept Connectivity Gateways • Built end to end FIX based gateways to establish reliable communication between trading systems and external markets - Creditex, Tradeweb, Bloomberg, GFI • Highly configurable gateways handle abrupt system shutdowns by syncing trades on a secondary process. Application Monitor • Implemented Heartbeat mechanism for gateways using Java threads • Developed an application to monitor gateway connectivity using TIBCO. • The software sends real-time updates to concerned teams. Made application configurable using spring and JMX Log file to XML test file Parsing • Developed a parser as a part of regression testing tool. Built and implemented logic to create XML test file from production logs. Graduate Analyst – Trade Processing Engine Dept Feed Validation Framework • Worked on Ruby framework to validate reports generated by trading system • The framework triggers report generation procedure if the report is invalid. Framework handles data constraints, size constraints, format constraints etc. TimesTen Cleaning and Update • Developed a logic to clean large volume duplicate entries in TimesTen, an in-memory database. Updated table Schema to resolve underlying bug