Skip to content
View srihasInaganti's full-sized avatar

Highlights

  • Pro

Block or report srihasInaganti

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
srihasInaganti/README.md

Hi, I'm Srihas Inaganti πŸ‘‹

CS + Applied Math @ University of Maryland

LinkedIn Email GitHub


🧭 About Me

  • πŸŽ“ B.S. Computer Science & Applied Mathematics, University of Maryland β€” Expected May 2028 (Dean's List, GPA 3.79)

🧱 Tech Stack

Languages

Java Python Go C Kotlin Swift JavaScript SQL HTML5 CSS3

Frameworks & Libraries

PyTorch React FastAPI Spring Boot Spring Data JPA

Cloud & DevOps

AWS Terraform Docker GitHub Actions

Data & Tools

PostgreSQL MongoDB SQLite DynamoDB Git Jira Bitbucket


πŸš€ Featured Projects

πŸ–ΌοΈ Serverless Image Processor

Live Code

AWS Β· GitHub Actions Β· Terraform

Event-driven pipeline decoupling ingestion from compute via SQS.

  • ⚑ 350+ req/sec sustained, 20ms p50 / 110ms p99
  • πŸ” Recovered 100/100 jobs from injected SIGKILL crashes, zero stuck/duplicate completions
  • πŸ”’ Blocked a verified zombie-worker race with a DynamoDB fencing token

πŸ—³οΈ Raft-KV

Code

Go

A from-scratch Raft consensus implementation with a linearizable KV store.

  • βœ… 107k+ commands validated across 6k chaos iterations, zero safety violations
  • ⏱️ 556ms median / 814ms p99 leader failover across 100 live kills (5-node cluster)
  • πŸ“– 7.6k reads/sec at 1.0ms latency via ReadIndex, bypassing Raft log disk commits

🧠 Mini-vLLM

Code

Python Β· PyTorch Β· torch.compile

LLM inference engine implementing PagedAttention and continuous batching.

  • 🎯 Reaches ~20% of vLLM's throughput
  • πŸ“ˆ 10.8x concurrency-32 throughput boost (106 β†’ 1,143 tok/s)
  • πŸ”Œ OpenAI-compatible REST API for drop-in client integration

⚑ OCR Fixture Code Detector

Live Code

Python Β· PyMuPDF Β· React Β· Document AI

Automates manual fixture-code review from electrical plans.

  • 🎯 88% detection accuracy across 23 real-world electrical plans
  • πŸ” Custom regex over raw OCR tokens to isolate true fixture codes from circuit numbers
  • πŸ–±οΈ Interactive bounding-box canvas with CSV export

πŸ“Š GitHub Stats


πŸ“« Let's connect: srihas.inaganti@gmail.com Β· LinkedIn

Pinned Loading

  1. Hack4Impact-UMD/hagar Hack4Impact-UMD/hagar Public

    TypeScript 11

  2. detextion detextion Public

  3. job-platform job-platform Public

    Python

  4. raft-kv raft-kv Public

    Go

  5. mini-vllm mini-vllm Public

    Python

  6. ocr-fixture-code-detector ocr-fixture-code-detector Public

    TypeScript