Vaibhav Bhandari

I build production-grade backend systems and LLM-powered applications, spanning APIs and services, data pipelines and integrations, retrieval systems, agent-driven workflows, and scalable infrastructure.Looking for my next role!

Resume GitHub LinkedIn Email

Featured Projects

SmartPatch: Intelligent Bug Repair

A pipeline for localizing and fixing bugs in software repositories using the SWEBench dataset.

PythonPyTorchTransformersvLLMQwen2.5-7BDeepSeekCoder-V2-InstructBM25MultiprocessingGit

Repo

ConverSync: Persistent LLM Chat with RAG + Caching

RAG powered chatbot with Response Time Optimizations

PythonGenAIRAGFAISSAWSFlaskRedisDynamoDB

Repo

AcademicAlly: Learning Enhancement Hub

AcademicAlly is a comprehensive web-based solution aimed at providing insightful feedback to students to improve their academic performances and aid faculties by giving them valuable inferences and pointers to review their teaching methods.

ReactJSMySQLFlaskGoogle Cloud PlatformHTMLCSS

Repo

TypeLock: Intelligent Keystroke-Based Security

Behavioral authentication system using keystroke dynamics and machine learning to verify user identity.

PythonFlaskJavaScriptSVM (scikit-learn)MongoDBHTML/CSS

Repo

PartSelectRAG: Appliance Parts Retrieval System

End-to-end retrieval-augmented system combining database queries with vector search for structured and unstructured product data.

PythonFastAPIPostgreSQL (RDS)PineconeReactViteSeleniumLLMs

Repo

MCP SQL Executor: Constrained LLM Database Agent

Backend service enabling safe, schema-aware SQL execution from LLM prompts using the Model Context Protocol (MCP).

PythonFastAPIPostgreSQL (RDS)MCPLLMsGit

Repo

Experience

Software Developer — Easley Dunn Productions Inc

Torrance, California

Aug 2025 — Present

Designed and implemented real-time multiplayer systems for a production racing game, handling authoritative state, synchronization, and event ordering across connected clients.
Built a dynamic leaderboard system to track race results and rankings in real time, ensuring consistency and correctness across client sessions.
Implemented client–server communication using RPCs and synchronized state updates, optimizing message flow to reduce latency and duplication issues.

Software Developer — Amadeus

Bangalore, India (UK Timezone)

Jul 2021 — Dec 2023

Architected and implemented scalable services within a Passenger Processing Platform, handling 2M+ daily API calls (~100 TPS peak) powering core workflows like check-in, baggage handling, and boarding.
Resolved 100+ production and performance-related tickets, directly contributing to 99.9%+ system uptime and ensuring uninterrupted passenger services across airline operations.
Spearheaded the adoption of a Behavior-Driven Development (BDD) testing framework to enhance collaboration between developers, testers, and business stakeholders, reducing communication cycles by over 30%.

Intern — Amadeus

Bangalore, India

Jan 2021 — Jun 2021

Developed a full-stack Django–Angular application to automate Inter-Airline Message retrieval, reducing manual processing time by over 80% and streamlining airline partner workflows.
Enhanced codebase maintainability through code modernization efforts including migrating 400+ regression tests from Perl to PyTest and resolving 20+ cyclic dependency conflicts, reducing build times by approximately 30%.

Intern — Samsung Research Institute

Bangalore, India

Jan 2020 — May 2020

Built an end-to-end Neural Machine Translation (NMT) pipeline for English → Indic languages using seq2seq architectures (BiLSTM/Transformer), covering data preprocessing, tokenization (BPE/SentencePiece), model training, and evaluation.
Optimized training pipelines through batching, earlystopping, and GPU-accelerated preprocessing, reducing model training time by ~25%.