Shaurya Bajpai
Full-Stack Software Engineer / Applied AI Developer. Open to internships and full-time 2027 software engineering roles, and happy to relocate anywhere in India.
About
I am a full-stack software engineer studying B.Tech Computer Science at LNMIIT, Jaipur, graduating in 2027. I have interned as a Software Development Engineer at Marine Edge and as a Frontend Web Developer at APJ Academy, working on platforms that serve thousands of users. Outside of internships I build and deploy my own projects: an AI stock assistant with a tool-calling agent, an emergency response platform that won a hackathon runner-up prize, a microservice e-commerce backend, and a decision engine I deliberately built without an LLM after analysing where a rule engine was safer. I work comfortably from the database layer to the UI, and I care about correctness, honest evaluation, and readable code. I have solved more than 300 data structures and algorithms problems in C++.
Experience
Software Development Engineer Intern, Marine Edge (March 2025 - January 2026)
- Built and maintained backend services and REST APIs for a web platform serving more than 7,000 users.
- Integrated the Razorpay REST API, handling authentication, data mapping, and error and retry paths.
- Implemented a role-based access control system with defined roles and permissions.
- Designed relational schema on PostgreSQL and optimized queries and indexes, contributing to a 30 percent faster load.
- Automated deployment with a Git-based workflow and traced production issues to root cause through application logs.
Frontend Web Developer Intern, APJ Academy (August 2024 - February 2025)
- Built responsive, cross-device user interfaces in React.js from project requirements.
- Optimized database queries, reducing data retrieval latency by 25 percent through query and index analysis.
- Implemented real-time data synchronization over WebSockets for low-latency updates.
- Collaborated with the team on code reviews and code quality.
Projects
- OARFIN (live): Emergency response platform: scrapes disaster news with Playwright, filters each item through a Gemini yes/no relevance classifier, and routes people to the nearest shelter by merging Overpass and FEMA data with failover. Runner-up at LNMIIT HackCrux.
- NammaStocks (live): Stock analytics web app with Stockie, an AI assistant that acts through tool calls and answers from live market data. I built the tool-call loop, token-by-token SSE streaming, a retrieval-augmented prompt, and a pluggable OpenAI / Anthropic / Gemini / Ollama provider layer.
- Ecom Microservice GraphQL (live): Three Docker services behind one GraphQL API, communicating over Kafka topics, with a React storefront. Debugged a silent cross-service message loss down to mismatched producer and consumer topic names.
- Vera, but Better: A deterministic decision engine that scores inputs across 26 weighted categories into one ranked outcome. I chose a rule engine over an LLM on purpose after studying the failure modes, then validated it against 100 scenarios.
- DevLab: A dual CI/CD sandbox: the same change builds, tests, and deploys through both GitHub Actions and Jenkins, with Bash setup scripts and Docker Compose for reproducible environments across machines.
- TorrentEdge: A distributed-systems study: a Kafka message queue decoupling ingestion from processing, Nginx as a reverse proxy for HTTP, WebSocket and inter-service traffic, and one-command Docker Compose startup. The peer-protocol layer is scaffolded.
- ReconIQ (live): A reconciliation agent that matches orders, gateway settlements and bank lines. Deterministic passes handle exact, fuzzy, split (1:N) and batch (N:1) matches, and an LLM is used only for bank narrations the rules cannot read. Anything left over becomes a reason-coded exception.
- Payeazie (live): A payment operations app: asynchronous charge processing on Redis and BullMQ workers, a state-machine payment lifecycle with audit logging, refunds with eligibility rules, retry and stuck-payment recovery, and an internal dashboard for operators.
- Mini ATS (live): An applicant tracking system with a drag and drop hiring pipeline, stage rules enforced in both the UI and the API, SQL full-text search, CSV import and export, and a metrics dashboard. 25 tests, including the API against a real Postgres.
- ElevateBox Voice Agent: An outbound voice sales agent in Hindi, Telugu and English. Vapi runs the call with Deepgram speech and a Groq model, Gemini classifies each lead as Hot, Warm or Cold, and WhatsApp follow-ups go out through the Meta Cloud API with retries and a template fallback.
- Handsfree Media Controller: A Chrome extension that controls video with hand gestures from the webcam. MediaPipe hand tracking runs on the device, a debouncer stops false triggers, and every gesture can be remapped to play, mute, speed or skip.
Achievements
- Runner-Up, LNMIIT HackCrux Hackathon, for OARFIN.
- Finalist, HackRx 6.0 Hackathon, by Bajaj Finserv Health.
- Top 5 college teams shortlisted for the Smart India Hackathon.
- GenAI Workshop, LNMIIT, scored 100 out of 100.
- 300+ DSA problems solved in C++.
Education
The LNM Institute of Information Technology (LNMIIT), Jaipur. B.Tech, Computer Science and Engineering, August 2023 to August 2027.
Contact
Email bajpaishaurya2911@gmail.com. GitHub, LinkedIn, LeetCode, Resume (PDF).
