P

Prabhakar Singh

Software Engineer

San Francisco, California, United States13 yrs 6 mos experience
Highly StableAI Enabled

Key Highlights

  • Designed RAG-LLM framework improving spam detection accuracy.
  • Led fraud prevention initiatives saving over $1.5M annually.
  • Expert in cloud service integrations and CI/CD automation.
Stackforce AI infers this person is a SaaS-focused software engineer with expertise in fraud detection and cloud services.

Contact

Skills

Core Skills

Distributed SystemsAi/llm-based SolutionsFraud DetectionBackend DevelopmentCloud Service IntegrationsCi/cd AutomationCloud Service MigrationIntegration With Machine Learning ServicesSoftware DevelopmentSecurity Framework DevelopmentMobile Application DevelopmentAuthorization And Authentication

Other Skills

JavaPythonPayout control systemsReal-time fraud detectionPerformance toolingPerformance monitoringTestingDeploymentREST API designAudit loggingUser information synchronizationContinuous Integration and Continuous Delivery (CI/CD)gRPCData StructuresAlgorithms

About

Experienced software engineer with 13+ years of building scalable systems and leading impactful projects at Meta, DoorDash, and Oracle. I specialize in distributed systems, spam measurement, fraud detection infrastructure, and AI/LLM-based solutions, with hands-on expertise in Java, Kotlin, Python, and cloud-native development. At Meta, I designed a RAG-based LLM framework that improved spam detection accuracy across Instagram, Facebook, and Threads. At DoorDash, I led fraud prevention initiatives that saved over $1.5M annually, built payout control systems, and streamlined platform reliability. My work at Oracle focused on cloud service integrations, CI/CD automation, and performance tooling. I’m also a Fellow of IETE and a Senior IEEE member, contributing to global technical communities through research and conference leadership.

Experience

13 yrs 6 mos
Total Experience
3 yrs
Average Tenure
1 yr 6 mos
Current Experience

Meta

Software Engineer

Dec 2024Present · 1 yr 6 mos · Menlo Park, California, United States · Hybrid

  • Designed and implemented a RAG-LLM based solution to assess spam prevalence across Meta’s family of products, including Instagram, Facebook, and Threads.
  • Launched a Retrieval-Augmented Generation (RAG) system to measure spam on Instagram, achieving a 90% F-score, a significant improvement over the 80% F-score of the standard LLM.
  • Optimized the RAG prompt generator async job, reducing Instagram sample retrieval errors by 94% and enhancing retrieval accuracy.
  • Integrated confidence scores for spam labels in the RAG database, enabling high-quality labels for LLM inference, fine-tuning spam detection models, and auto-refreshing ML models for continuous improvement.
JavaPythonAI/LLM-based solutionsDistributed systems

Doordash

Software Engineer

Feb 2022Dec 2024 · 2 yrs 10 mos · San Francisco Bay Area · Hybrid

  • 1. Payout Controls for Dasher Delivery Pay
  • Payout Deferrals – As a Tech Lead collaborating with a team of five engineers, I designed, developed, and launched a delivery-level payout deferral infrastructure for Dasher shift pay and anomalous pay, reducing fraud-related costs by $400K annually. I led rule requirement discussions and worked closely with Strategy & Ops partners and backend engineers enhance real-time fraud detection.
  • Line-Item Payout Block on Deactivation – Designed and launched a payout blocking mechanism for Dashers upon deactivation to mitigate legal risks, saving $100K annually by preventing fraudulent payouts. Aligned stakeholders from legal, finance, compliance, tax, and DE teams on this initiative.
  • 2. Dasher Account Locking
  • Designed and implemented the Dasher Account Locking feature and fraud detection rules, resulting in $800K in recurring annual savings and reducing false positive deactivations by 10x, thereby minimizing churn of legitimate Dashers.
  • 3. Unified Contract Violations (UCV)
  • Designed the UCV framework to encourage better Dasher behavior, reduce recidivism, and enhance transparency around contract violations. Enabled an appeals process to provide Dashers with due process.
  • 4. Delaying Dasher Deactivation
  • Built infrastructure to delay deactivation for Dashers actively completing deliveries, reducing lost visibility costs by $30K per month for Parcel orders.
  • 5. Real-Time Suspicious Deliveries Support
  • Implemented backend support for flagging suspicious deliveries and enabling line-item payout blocks.
  • Improved fraud agents' efficiency by allowing them to focus on high-risk cases.
  • 6. Tech Stack Migration
  • Extracted the Dasher deactivation API from the legacy stack, reducing deactivation error rates from 34% to 2%.
  • Migrated fraud detection rate limiter rules to a new infrastructure, achieving $150K in annual fraud cost savings and reducing detection latency from 400ms to 24ms, enabling faster fraud decision-making.
Fraud detectionPayout control systemsReal-time fraud detectionBackend development

Oracle

2 roles

Principal Software Engineer

Oct 2019Feb 2022 · 2 yrs 4 mos

  • Automated Sauce video platform deployment with Oracle Cloud Infrastructure using Shepherd and Terraform providers.
  • Implemented Oracle Cloud Infrastructure console integration for Oracle Content Management (OCM) that provides easy interface for customers to launch OCM service within few minutes.
  • Worked on Service platform integration with Oracle Content Management product in order to abstract out features like AuthN, AuthZ, Auditing, Limits enforcement, and other platform bar items from the service developer.
Cloud service integrationsCI/CD automationPerformance tooling

Senior Member of Technical Staff

Feb 2016Sep 2019 · 3 yrs 7 mos

  • Migrated Oracle Content and Experience Cloud Service (CEC) from Internal Compute to External Compute.
  • Worked on end to end integration of CEC with Oracle Machine Leaning service running on Oracle Cloud Infrastructure Container Engine for Kubernetes.
  • Integrated Metering and Quota service with CEC cloud which allows dynamic resource usage reporting. Customer will pay only for what they use even they subscribe for higher quota.
  • Implemented WebCenter Portal Performance monitoring tool using Oracle JET. This tool helps diagnose performance issues with oracle software at each stage of product lifecycle.
Cloud service migrationIntegration with Machine Learning servicesPerformance monitoring

Amazon

SDE Intern

May 2015Aug 2015 · 3 mos · Seattle, Washington

  • Over the course of 12 weeks , I got opportunity to work on three different projects. I was responsible for end to end delivery of each project which included writing design document, implementing feature, testing and deployment to production.
  • Below is the brief description of the projects :
  • Implemented Vendor specific redemption instruction support in Digital Delivery confirmation emails. It will reduce customer service contacts related to product redemption by 2% and increase 1-year ops by 1MM+.
  • Added Before You Buy widget for courseware products to better serve students who purchase courseware. This widget will lead to a better converting detail page, less customer issue related calls and less purchase cancellations.
  • Implemented Amazon Instant Access Integration test suite for Vendors which provides better confidence to Vendors on their Instant Access system, reduces development time and saves Amazon resources.
Software developmentTestingDeployment

Oracle server technologies

2 roles

Senior Member Technical Staff

Promoted

Oct 2013Jul 2014 · 9 mos · Bengaluru, Karnataka, India

  • As being senior software developer in security team of Oracle WebCenter Portal, some of my work included :
  • Designing and developing multirole support for users and enterprise groups in WebCenter Portal
  • Designing and implementing enterprise group addition, deletion and listing through REST api’s in WebCenter Portal
  • Developing ADF mobile application to showcase side by side integration of WebCenter Sites and Portal
Security framework developmentREST API designMobile application development

Member Technical Staff

Jun 2011Sep 2013 · 2 yrs 3 mos · Bengaluru, Karnataka, India

  • I was in security team of Oracle WebCenter Portal product. I was responsible for handling the entire security framework that the Portal depended upon to make all kinds of authorization and authentication decisions.
  • Some of highlights of my work here include --
  • Enhanced Weblogic scripting tool for synchronizing user information with Oracle Identity Store in WebCenter Portal application
  • Designed and implemented Oracle FMW audit logging service integration with WebCenter Portal
Authorization and authenticationAudit loggingUser information synchronization

Education

University of Southern California

Master of Science (M.S.) — Computer Science

Jan 2014Jan 2016

Motilal Nehru National Institute Of Technology

Bachelor of Technology (B.Tech.) — Computer Science and Engineering

Jan 2007Jan 2011

Stackforce found 100+ more professionals with Distributed Systems & Ai/llm-based Solutions

Explore similar profiles based on matching skills and experience