Architecting
Intelligent Logic.
Specializing in enterprise backend systems, Python APIs, and autonomous AI-driven code analysis. Transforming structured data and complex architectures into scalable software solutions.
VARUN PARUCHURI
I'm a Software Engineer completing my B.Tech in Computer Science and Engineering at Malla Reddy University. I build backend systems and REST APIs in Python using FastAPI and Flask, and I've shipped full-stack projects end to end — most recently AetherOps, an AI-powered codebase analyzer built with Next.js and Gemini that automatically detects and fixes code issues. I'm comfortable across the stack, from React on the frontend to PostgreSQL on the backend.
Experience
Data Science / AI-ML Intern
Aug 2025 - Sep 2025- ▹ Designed and optimized ML preprocessing pipelines in Python, boosting overall model accuracy and efficiency.
- ▹ Performed data cleaning and feature engineering to extract actionable insights from structured datasets.
- ▹ Automated validation tasks with scripts, cutting manual effort by 30% and ensuring strict data quality.
Featured Projects
AetherOps: Real-Time SRE Codebase Analyzer
Architected an enterprise-grade SRE codebase analyzer using Next.js, TypeScript, and Drizzle ORM with Neon PostgreSQL to monitor repository health in real time. Built an autonomous AI code-review pipeline (Gemini 1.5 Pro) that scans structures, flags vulnerabilities, and renders side-by-side auto-remediations at 60 FPS.
Resume Screening & ATS Match Engine
- ▹ Built a resume parsing and deduplication service in Python/FastAPI that extracts candidate data from PDF/TXT, flagging duplicates via SHA-256 hashing and fuzzy scoring (>88% threshold).
- ▹ Engineered a dual-mode scoring engine routing to Groq (LLaMA 3.3 70B) or GPT-4o-mini for JD matching, with local offline heuristic fallbacks.
- ▹ Deployed as a serverless FastAPI app on Vercel utilizing `/tmp` directory architecture for real-time asset processing.
AI-Powered Job Search Automation Pipeline
Built two scheduled workflows on Make.com integrating SerpApi, Groq (LLaMA 3.3 70B), Google Sheets, and Telegram Bot API to source, deduplicate, and screen job postings.
Used an LLM to score each posting against a target candidate profile and automatically generate highly tailored cover letters, drastically reducing manual search effort.
Built a second AI workflow to classify referral-offer posts, filtering spam from genuine networking opportunities and pushing verified matches via instant Telegram alerts.
Technical Architecture & Skills
Hover over category cubes to engage interactive 3D magnetic tracking and highlight core engineering competencies.
Backend & APIs
Server Architecture
AI & Machine Learning
Intelligence Layer
Frontend & Web
Client Interface
DevOps & Tools
Pipeline & OS