Software Engineer • AI/ML Systems • Graduate CS Student

Vishal Vilas Shinde

Graduate computer science student at California State University, Fullerton with hands-on experience in backend engineering, machine learning pipelines, and full-stack product development. I build reliable systems with measurable performance gains, from CI/CD-driven teaching workflows to production-ready APIs and retrieval-based AI applications.

Based in California, United States

Email itsvishal2417@gmail.com

Resume View Projects

Experience

Engineering, research, and teaching work

Experience across teaching infrastructure, computer vision research, backend systems, and full-stack product engineering with a focus on performance, reliability, and measurable outcomes.

Teaching Assistant

California State University, Fullerton

California, USA

Aug 2025 - Present

  • Responsible for holding office hours and conducting lab sessions for CPSC 223P: Python Programming.
  • Built a CI/CD pipeline with automated unit testing and pull-request-based workflows, reducing manual grading time by 80%.

Research Assistant

California State University, Fullerton

California, USA

Jun 2025 - Present

  • Engineered a PyTorch computer vision pipeline for 5-class diabetic retinopathy classification using ResNet-50 with two-phase fine-tuning.
  • Achieved 91% accuracy and QWK 0.9167, and used Grad-CAM to guide error analysis.
  • Introduced object-aware Neural Style Transfer with a curated style bank to reduce overfitting and domain shift, targeting improved QWK and >95% accuracy.

Software Engineering Intern

Digital Product School by UnternehmerTUM

Munich, Germany

May 2023 - Jul 2023

  • Developed a backend API using Java (Spring Boot), PostgreSQL, and AWS App Runner.
  • Implemented indexing, table partitioning, and asynchronous processing for stable performance under heavy traffic.
  • Improved a scikit-learn anomaly detection pipeline by 15% through feature engineering and radar sensor data collection in Python.

Full-stack Developer Intern

Mezchip

Bengaluru, India

Jul 2022 - Oct 2022

  • Built a real-time chatbot using Flask, Preact, and PostgreSQL, serving 1,000+ users, achieving sub-200ms response times and decreasing login failures by 3% after implementing JWT-based authentication.
  • Optimized package dependencies to reduce the chatbot's bundle size by 15% and implemented 8 secure RESTful APIs with OAuth2 authentication, rate limiting, and input validation to enhance performance and security.
  • Implemented a debounced autocomplete search feature, reducing API calls by 40% and improving search responsiveness.

Projects

Selected work

A mix of LLM tooling, cloud-native RAG systems, developer infrastructure, and real-time communication experiments built for practical use and clear engineering tradeoffs.

SafePrompt

Fine-tuned an LLM with QLoRA for specialized PII redaction, producing lightweight LoRA adapters that generate structured placeholder-based redactions.

  • Micro-F1 around 90% with span-level evaluation pipeline
  • CPU-only FastAPI service for offline inference on commodity hardware
QLoRA Llama 3.2 FastAPI Web Extension Hugging Face
Model Adapter Repository

CourseHero

Built and deployed a Python RAG API on Cloudflare Workers using Workers AI and Vectorize, with a validated JSON contract and safe rollout workflows.

  • Deterministic ingestion pipeline from CSV to strict NDJSON metadata
  • Rate limiting, observability, and staging/production wrangler environments
RAG Llama 3.1 Cloudflare Workers Vectorize Workers AI
Try Repository

DBsnap

Developed a unified backup and restore platform for four database engines behind a single CLI, packaged in a containerized runtime for portable execution.

  • AWS S3 cloud backup workflow with direct restore options
  • Engine-agnostic CLI abstractions for developer and server environments
Python CLI Docker AWS S3 Boto3
Docker Image Request GitHub

Learn WebRTC

Built a full-stack WebRTC playground with custom signaling for real-time video, audio, and data channel communication, plus an interactive debugging assistant.

  • Custom Node.js / Socket.io signaling server and RTCPeerConnection flows
  • Gemini-powered debugging assistant for SDP, signaling, and ICE analysis
TypeScript WebRTC Next.js Socket.io Google Gemini
Try Repository
View More

Contact

Let’s build something useful

I’m open to Applied AI and backend internship opportunities. Reach out for collaboration, internships, or full-time roles.