Balvinder Singh

Software Engineering Intern, Backend Systems & Applied AI

6-month internship from May 2027. MSc AI, CentraleSupélec (graduating Dec. 2027).

Paris, Francesbalvinder.fr@gmail.comlinkedin.com/in/balvind3rsinghgithub.com/balv1nder

MSc Artificial Intelligence student at CentraleSupélec with 1.5+ years at OYO, building backend services and internal tools for hotels in Europe, UK, USA and the UAE. Experienced in managing scalable distributed backend systems and co-author of two IEEE papers applying LLMs, retrieval-augmented generation (RAG) and machine learning to healthcare and cybersecurity. Looking for a 5–6 month internship from May 2027 to build reliable, scalable software and AI-powered features.

Experience

Associate Software Development Engineer

OYO (Oravel Stays), Gurugram, India

Aug. 2025 – Sep. 2026

  • Designed and deployed backend REST APIs for automated smart-lock and room-upsell products in the UK and UAE, streamlining check-in and contributing to a 15% increase in property revenue.
  • Built an end-to-end digital check-in system that replaces front-desk check-in with a self-service guest flow; projected operational savings of over $1M per year.
  • Developed in-house dashboards for the Motel 6 management system, reducing operational costs and improving reliability; team Building Block Award for supporting the Motel 6 (USA) acquisition, and handling critical on-call issues.

Software Engineering Intern

OYO (Oravel Stays), Gurugram, India

Jan. 2025 – Jul. 2025

  • Led projects to streamline hotel onboarding and financial contracting on the OYO 360 platform for 1,000+ hotels across Europe (Belvilla, DanCenter, CheckMyGuest); improved operational efficiency by 30%.
  • Owned month-end closure and year-end audit processes for merchant payouts across several countries, ensuring accurate settlement and reconciliation.

Information Technology Intern

Mahindra Logistics (Mahindra Group), Mumbai, India

Feb. 2024 – Apr. 2024

  • Cut API response time from 50+ s to ~500 ms (100× faster) on the LogiPulse KPI-monitoring system by adding Redis caching, optimising SQL queries and redesigning APIs.
  • Redesigned the operator interface, reducing task time by ~80%; 1st place among 200+ teams at the National Catapult Techathon for KPI-monitoring workflows used across 1,000+ warehouses.

Projects & Publications

AgentAudit: auditable LLM agent for financial queries

Java, Spring Boot, Spring AI, PostgreSQL, OpenTelemetry, Ollama

Oct. 2026

  • A Spring AI agent over a locally hosted LLM that answers portfolio-risk questions by invoking Java tools, with every run persisted to PostgreSQL as a session plus an ordered event log of tool calls, arguments and results, each correlated to its OpenTelemetry trace.
  • Caught the model asserting EUR 12M was higher than EUR 14M, reproducibly at temperature 0; the record isolated the fault to narration, not data.
  • Measured a prompt-based fix over three runs per arm: it did not correct the answer and added a tool round, visible only in the audit trail.

Pulse: distributed, rate-limited API service

Go, Redis, Docker, Prometheus, k6

Oct. 2026

  • Built a horizontally scaled Go service: custom round-robin load balancer with health checks, three stateless API replicas, a distributed rate limiter (atomic Redis Lua script, verified with the race detector) and a cache-aside layer; Prometheus metrics and Docker Compose.
  • Load-tested with k6: p99 of 8 ms at 1,500 req/s; identified the load balancer as the shared bottleneck and diagnosed a cascading failure caused by health-check timeouts under overload.

LLM-based clinical decision support

Google AI for Impact Finalist. IEEE ICDLT (paper presented)

Nov. 2025

  • Built a hybrid RAG pipeline (AWS Textract OCR → BioBERT embeddings in Pinecone → Gemini 1.5) for differential-diagnosis support, with dual-persona prompting and rolling summarisation to fit long patient histories in context.

Botnet attack detection for IoT devices

IEEE TENSYMP (paper presented)

Sep. 2024

  • Stacking ensemble with time-window feature engineering on N-BaIoT to classify Mirai and BASHLITE attacks on low-resource devices; 99.98% binary accuracy.

Education

MSc in Artificial Intelligence (M2, 60 ECTS)

CentraleSupélec, Université Paris-Saclay

Sep. 2026 – Dec. 2027

  • Big data for AI, multi-agent systems, deep learning, trustworthy and explainable AI.

B.Tech in Computer Science & Engineering (AI)

Netaji Subhas University of Technology

Dec. 2021 – Dec. 2025

  • CGPA 7.58/10. Coursework: Distributed Computing, Operating Systems, Database Management Systems, Data Structures & Algorithms.

Skills

Languages
Java, Python, C++, SQL, Go, JavaScript
Backend & data
REST APIs, Microservices, Spring Boot, ActiveMQ, Redis, PostgreSQL, MySQL, MongoDB
Infra & observability
Docker, Kubernetes, AWS, Prometheus, Git
AI / ML
Machine learning, Deep Learning, NLP, LLMs, RAG, Pinecone; AI Coding Assistants
Other
English C1 (IELTS 8.0), Hindi native, German A1. LeetCode 1600+. Top 10, Google APAC AI for Impact Hackathon