I'm Vishnu Vivek Valeti, a graduate student in Computer Science at the University of Florida. I build software that is scalable, performant, and production-ready, from clean React interfaces and concurrent Go backends to ML training pipelines on AWS.
Currently, I'm a Software Engineer Intern at ReplyQuick LLC, where I build scalable training pipelines processing 5,000+ dental images, develop PyTorch evaluation utilities, and integrate ML inference outputs with backend APIs using versioned schemas. Previously at UF, I built FastAPI-based backends containerized with Docker, reducing result-serving latency from 900ms to 280ms. I also built extensions for Droppy (macOS) that have accumulated 2,000+ downloads combined. I ship measurable outcomes, collaborate across engineering teams, and improve systems with a calm, methodical approach.
What I work with
University of Florida · Gainesville, FL
CourseworkAdvanced Data Structures, Distributed Operating Systems, Software Engineering, Natural Language Processing, Analysis of Algorithms, Distributed Multimedia Systems
Vignan University · Guntur, India
CourseworkData Structures, Object-Oriented Programming, Database Management Systems, Operating Systems, Machine Learning, Cloud Computing
ReplyQuick LLC
Shipped 8 versioned REST API endpoints in FastAPI with typed Pydantic schemas, replacing fragile client-side post-processing logic by moving validation to the server layer so client teams stopped patching schema mismatches in their integrations. Standardized service deployments with environment-based configuration, centralized logging, and CI/CD-ready artifacts, cutting new environment setup time from 2 hours to under 30 minutes. Built a Python data ingestion pipeline that processed 5,000+ dental images with schema validation and structured error reporting that surfaced failed records with full diagnostic context. Developed PostgreSQL-backed metrics storage and comparison reports for image-processing pipeline runs, reducing reviewer analysis time by 30% and enabling run-over-run quality tracking.
University of Florida
Engineered a data pipeline on AWS (S3, EC2) with PostgreSQL to manage structured research datasets, cutting query retrieval time by 35% through composite indexing and table partitioning. Containerized REST APIs in Python (FastAPI) with Docker, reducing p95 response latency from 900ms to 280ms via query optimization and connection pooling. Built a React/FastAPI analytics dashboard with reusable data-visualization components for dataset size, labels, and ingestion status, cutting weekly research reporting overhead by 30%.
Vignan University
Delivered attendance and staff-management features for a full-stack React/Node.js/MySQL academic portal serving 2,500+ daily active users, resolving open bug tickets across the attendance and staff modules over a 5-month engagement. Automated batch attendance workflows (mark-all, quick-edit), saving ~40 hours per week across 5 departments by eliminating repetitive per-student actions. Optimized slow SQL JOIN queries via MySQL indexing and query refactoring, reducing page load time from 3.2s to 1.0s. Implemented role-based access control (RBAC) and OAuth 2.0 authentication on grade and results endpoints, blocking unauthorized access paths flagged during pre-launch security testing.
Rule-based release go / no-go on every PR
LLM-powered API regression testing platform. Uses the Claude API to generate adversarial test scenarios, runs spike/soak/stress load profiles, and a GitPython diff engine classifies breaking changes across 6 impact tiers — covered by 560 pytest tests.
Python · FastAPI · Claude API · PostgreSQL · React · WebSocket
99.7% success · 45ms p95 · at-least-once delivery
Event-driven webhook delivery service in async Python/FastAPI processing ~1,200 events/sec via Redis-queued Celery workers, with jittered exponential backoff and idempotency keys for at-least-once delivery. Prometheus-instrumented success rate, queue depth, and p95 latency; PostgreSQL-backed delivery state.
Python · FastAPI · Celery · Redis · PostgreSQL · Prometheus
Sub-10ms BM25 search · 4,300 files/sec indexing
Parallel C++17 code search engine indexing ~4,300 files/sec into a SQLite WAL database via std::thread workers, batched writes, and a std::filesystem file walker. Serves Okapi BM25 full-text search and trigram fuzzy filename matching at sub-10ms latency. CI on GCC, Clang, and MSVC.
C++17 · SQLite · CMake · BM25 · Multithreading · React
Auth, subscriptions, and career-site scraping end-to-end
Full-stack job-tracking platform: Go + Gorilla Mux backend, React frontend, JWT auth with email-OTP password reset, and per-user company subscriptions that scrape career sites for new-grad SWE listings.
Go · React · PostgreSQL · JWT · Cypress · Docker
Actor-model concurrency + rate limiting in pure Go
Concurrent Go backend with in-memory state — sync.Map registries, per-entity RWMutex, actor-style Receive() dispatcher, token-bucket rate limiting, atomic counters with CAS, and a live /metrics endpoint.
Go · Actor Model · sync.Map · Token Bucket · REST
Hot-reloadable rules over native WebSockets
Real-time chatbot with hot-reloadable JSON rule evaluation, native WebSocket messaging, and containerized deployment.
React · TypeScript · Node.js · WebSockets · Docker
Offline-first Android, no telemetry
Privacy-focused Android habit tracker with streak tracking, smart reminders, and offline-first architecture using Material 3.
Kotlin · Jetpack Compose · Room · Hilt · Android
Signed R2 uploads + role-gated reader, all free-tier
Mobile-first manga reader SPA on Vercel with Cloudflare R2 image hosting. Secure signed-URL admin uploads, role-gated routes, lazy-loaded chapter pages, and bookmark/history persistence — all on free-tier infra.
React · TypeScript · Tailwind · Vite · Vercel · Cloudflare R2
20-week roadmap with offline sync, auth, and RLS
Next.js 14 study tracker for the JLPT N5 Japanese exam. 1,400+ tasks across a 20-week roadmap, Supabase auth with Row-Level Security, an offline sync queue, and Recharts progress analytics.
Next.js 14 · TypeScript · Supabase · Zustand · Recharts · Tailwind
I'm actively looking for full-time Software Engineering roles. If you think I'd be a good fit for your team, I'd love to hear from you.