San Francisco, CA

👋🏼 Hello I'mPavan Suresh

Building scalable backend systems and cloud infrastructure.

View Projects Download Resume

01. About

About Me

Pavan Suresh

Software Engineer

🎓3.92 GPA

☁️AWS Certified DVA-C02

Background

I'm a Software Engineer with 5 years of experience building scalable backend systems, AI-powered platforms, and cloud-native applications. My core strengths are Python, FastAPI, REST APIs, microservices, AWS, GCP, Docker, Kubernetes, and distributed systems. I recently completed my M.S. in Information Systems from Saint Louis University with a 3.92 GPA, and I’m also an AWS Certified Developer – Associate.

My recent work focuses on LLM-based systems, RAG pipelines, semantic search, Kafka-based event processing, Redis caching, and high-performance API design. At Shopify, I worked on AI and ML infrastructure where I improved response accuracy, reduced inference latency, optimized LLM costs, and built backend services that supported thousands of active merchants.

Before that, at Accenture, I built and optimized large-scale data engineering workflows using Python, PySpark, SQL, Airflow, Spark, AWS S3, Glue, EMR, Lambda, and SQS. Across my projects, I focus on clean architecture, secure cloud design, observability, automation, and building systems that are reliable, maintainable, and easy to scale.

02. Projects

Featured Projects

01 Solo Build · Cloud Architecture

AWS Document Processing Pipeline

Python FastAPI AWS S3 SQS SNS EC2 RDS Docker IAM VPC CloudWatch

A production-style asynchronous document processing pipeline built entirely on AWS. Users upload PDF, CSV, or image files via a FastAPI REST endpoint - files are immediately stored in S3 and a processing job is enqueued in SQS, returning 202 Accepted without ever blocking the request thread. An EC2 worker in a private subnet polls the queue, processes the document, writes results back to S3, tracks job status in PostgreSQL (RDS), and fires an SNS email notification on completion.

Key engineering decisions: zero credentials on servers (EC2 uses IAM roles exclusively), full VPC isolation with public/private subnets, Dead Letter Queue preserving failed messages after 3 retries, structured JSON logging queryable in CloudWatch, and horizontal scaling by running multiple worker containers against the same SQS queue.

Async 202 Pattern DLQ Handling IAM Role Auth VPC Isolation CloudWatch Monitoring Horizontal Scaling Zero Credential Design

GitHub

AWS Document Processing Pipeline Architecture

System Architecture

02 Lead Developer · GPT Integration

ForgeEd - AI-Enhanced Learning Management System

Python Flask OpenAI API SQLite JavaScript Chart.js Tailwind CSS

A full-stack AI-enhanced Learning Management System designed for university students, built over a 16-week Master's research project. The platform unifies academics, wellbeing, and career development into a single interface - solving the fragmentation problem of existing LMS tools like Canvas and Blackboard.

Core modules: SLU GPT (24/7 AI tutoring assistant powered by OpenAI GPT API), a 90-day career learning roadmap with day-locked progression, a dual quiz system with AI auto-scoring, daily wellbeing tracking across 5 metrics, and a predictive analytics dashboard computing GPA risk levels and personalized recommendations.

GPT-Powered Tutoring 90-Day Roadmap Dual Quiz Engine Wellbeing Tracking Risk Prediction Admin Dashboard 300+ Users

GitHub

ForgeEd Business and Application Architecture

System Architecture

03 Solo Build · Full Stack

Blog Website

Python Flask SQLAlchemy Flask-Login CKEditor Bootstrap 5 PostgreSQL

A full-featured multi-user blog platform built with Flask and Python. Role-based access control gives admins full CRUD over posts via a CKEditor rich text editor, while registered users can read and comment. Authentication uses Werkzeug-hashed passwords with Flask-Login session management, Gravatar auto-generates profile avatars, and the data layer switches between SQLite in development and PostgreSQL in production via environment-variable config - with a Heroku-ready Procfile included.

Role-Based Access CKEditor Rich Text Hashed Auth Gravatar Avatars SQLite + PostgreSQL Heroku-Ready

GitHub

03. Experience

Work Experience

Shopify Oct 2024 – Present

Software Engineer – Generative AI & ML Infrastructure

San Francisco, CA

Improved AI response accuracy by 32% by implementing Python-based RAG pipelines with contextual merchant data, optimizing prompt engineering and semantic re-ranking
Reduced inference latency by 41% by designing asynchronous FastAPI services with request batching, Redis caching, and API Gateway integration
Decreased infrastructure cost by 28% via dynamic batching and response caching, supporting AI-driven workflows used by 3,000+ active merchants daily
Built LLM-powered APIs for merchant workflows using FastAPI, LangChain, and OpenAI APIs with React.js frontend integration
Developed retrieval pipelines using FAISS, embeddings, and hybrid search combining semantic similarity with keyword ranking
Designed agent orchestration systems with tool-calling capabilities for product updates, campaign creation, and analytics insights
Implemented event-driven pipelines using Kafka and async workers on GCP with Redis pub/sub for real-time processing
Deployed containerized services using Docker and Kubernetes on GCP with blue-green deployment strategies

Accenture Sep 2020 – Dec 2023

Software Engineer

Bangalore, India

Migrated legacy SAS-based workflows to Python and PySpark pipelines, reducing data processing latency by 38%
Optimized distributed Spark jobs through partition tuning and caching, reducing execution time by 42%
Built scalable ETL pipelines using Python, PySpark, and SQL integrated with Airflow scheduling and batch workflows
Deployed data pipelines on AWS using S3, Glue, EMR, Lambda, and SQS for event-driven processing
Implemented Kafka and Spark Streaming on AWS for real-time ingestion across microservices-based data systems
Built FastAPI-based REST APIs on AWS behind API Gateway with JWT authentication and rate limiting
Containerized data processing services using Docker and Kubernetes with auto-scaling in production

04. Skills