Prabhakar Prasad

Software Engineer building scalable distributed systems, AI-powered automation platforms, and real-time trading infrastructure. Previously SDE II at Amazon.

Email GitHub LinkedIn
Prabhakar Prasad

About

I am a full-stack engineer with experience building high-scale backend systems (10M+ TPS), AI-enabled workflows, and cloud-native platforms on AWS. I enjoy owning problems end-to-end, designing clean architectures, and shipping systems that are reliable, observable, and easy to extend.

Experience

Amazon — Software Development Engineer II

2022 – 2024 · Bangalore, India

  • Built AI-powered ticket resolution and audit platforms using LLMs and event-driven AWS architecture.
  • Improved audit accuracy from 88% to 97% using automated anomaly detection pipelines.
  • Designed secure, scalable internal tools with modern authorization and observability.

Amazon — Software Development Engineer I

2020 – 2022 · Bangalore, India

  • Designed and scaled high-throughput microservices with <100ms p99 latency.
  • Migrated legacy systems to AWS, reducing infrastructure costs by 35%.
  • Maintained 99.95% uptime during peak traffic events like Prime Day.

Projects

Algorithmic Trading Platform

End-to-end trading system for Indian markets with real-time data ingestion, multi-timeframe analysis, strategy orchestration, and live dashboards.

Argus – Eye Health Screening Platform

Workflow automation platform for clinical eye exams, reducing patient processing time by 50%. Winner of TopCoder Hackathon.

Clickbait Detection Chrome Extension

ML-powered browser extension for real-time headline classification with 91% accuracy.

Projects – Architecture Deep Dive

Algorithmic Trading Platform – System Architecture

Designed as a low-latency, horizontally scalable system with clear separation between ingestion, analysis, execution, and presentation layers. 

┌───────────────┐        ┌───────────────────┐
│ Broker APIs   │──────▶ │ Market Data Fetch │
│ (Fyers, etc.) │        │ 5m / Tick Feeds   │
└───────────────┘        └─────────┬─────────┘
                                    │
                                    ▼
                          ┌───────────────────┐
                          │ Celery Workers    │
                          │ 990+ parallel     │
                          │ analysis tasks    │
                          └─────────┬─────────┘
                                    │
        ┌───────────────────────────┼───────────────────────────┐
        ▼                           ▼                           ▼
┌───────────────┐        ┌───────────────────┐        ┌───────────────────┐
│ Redis Cache   │        │ PostgreSQL        │        │ Strategy Engine   │
│ Hot data TTL  │        │ Historical data   │        │ MTF / Signals     │
└───────────────┘        └───────────────────┘        └─────────┬─────────┘
                                                                      │
                                                                      ▼
                                                        ┌───────────────────┐
                                                        │ Django Channels   │
                                                        │ WebSocket layer   │
                                                        └─────────┬─────────┘
                                                                      │
                                                                      ▼
                                                        ┌───────────────────┐
                                                        │ React Dashboard   │
                                                        │ Live updates      │
                                                        └───────────────────┘
  • Latency-optimized real-time updates (<100ms) via WebSockets
  • Fault-tolerant async processing using Celery + Redis
  • Designed for strategy extensibility and broker abstraction

AI-Enabled Seller Ticket Resolution – Amazon

Internal platform enabling auditors to resolve complex seller escalations using AI-assisted multi-step workflows. 

┌───────────────┐
│ Seller Ticket │
└───────┬───────┘
        ▼
┌────────────────────┐
│ Event Ingestion    │
│ (SQS / SNS)        │
└─────────┬──────────┘
          ▼
┌────────────────────┐
│ LLM Orchestrator   │
│ Claude + Rules     │
└─────────┬──────────┘
          ▼
┌────────────────────┐        ┌────────────────────┐
│ Data Fetch Layer   │◀──────▶│ DynamoDB / Redshift│
└─────────┬──────────┘        └────────────────────┘
          ▼
┌────────────────────┐
│ Audit UI (Web)     │
│ Human-in-the-loop  │
└────────────────────┘
  • Agentic AI workflows with full auditability
  • Event-driven design for scale and traceability
  • Improved decision consistency and audit accuracy

Skills

Backend & Cloud

Python, Java, Django, Spring Boot, FastAPI, AWS, Microservices, Event-Driven Systems

Frontend

React, TypeScript, Angular, HTML, CSS, WebSockets

AI & Data

LLM Integration, LangChain, TensorFlow, Scikit-learn, Redis, PostgreSQL