Software engineering student working on backend systems, distributed applications, and AI/ML projects. I like understanding how systems work under the hood, and turning ideas into software that works reliably.
→Building Dockerized microservices at Mahindra Comviva
→Working on backend systems & REST APIs
→Exploring AI/ML applications
→Practicing DSA — 400+ LeetCode problems solved
Experience and Achievements
Professional Workspace
~/sidhant/experience/
01 / experience/
Experience
├── 01_comviva/
Software Development Intern — Mahindra Comviva· GurugramcurrentJun 2026 – Present
Developing a Dockerized, microservices-based proactive incident monitoring platform using Python, Flask, and Docker to detect service failures in distributed telecom systems.
Building REST APIs, health monitoring, runtime metrics, structured logging, and inter-service communication to simulate production-grade backend infrastructure.
Designing scalable backend components for automated incident detection and remediation workflows.
├── 02_ericsson/
Software Development Intern — Ericsson· NoidaJun 2025 – Jul 2025
Contributed to backend services for a digital payments platform, working on API development and service-layer logic.
Improved API response performance by identifying bottlenecks and optimizing request handling.
Gained hands-on experience with containerized environments using Docker and CI/CD workflows.
Dockerized microservices (OAM, BAC, Monitoring Engine) that proactively detect failures in distributed telecom systems.
REST APIs, health checks, runtime metrics, structured logging, and inter-service communication over Docker networking.
Modular design built for continuous monitoring and future incident-management workflows.
PythonFlaskDockerREST APIsMicroservices
IntelliDocAdobe Hackathon Finalist
AI-Powered PDF Structure & Relevance Extractor
Layout-aware PDF hierarchy reconstruction using font metadata and heuristics.
Semantic ranking of sections via NLP to extract persona-relevant content.
Fully offline execution, containerized with Docker for portability.
PythonNLPPyMuPDFDocker
Finance Data Processing Backend
Role-based financial records system
Backend system managing users and financial records with role-based access control.
JWT-based authentication supporting viewer, analyst, and admin roles.
REST APIs for CRUD and filtering, plus aggregation endpoints for dashboard insights.
Node.jsExpressMongoDBJWTRBAC
EcoSync2nd Place · AWS PartyRock
Smart Home Energy Management System
Prompt-based workflows on AWS PartyRock that analyze household energy usage and suggest appliance-level optimizations.
AWS PartyRockPrompt EngineeringAI
projects/ proactive_incident_monitor
Proactive Incident Monitor
Built during my internship at Mahindra Comviva · 2026
Distributed Monitoring Platform — Dockerized microservices built to detect service failures in distributed telecom systems proactively.
Overview
In a distributed system, one service can fail while the rest keep running. This platform continuously checks service health and runtime metrics so failures are caught early rather than discovered after they cascade.
How It Works
Three independent services — OAM, BAC, and a Monitoring Engine — each run in their own Docker container.
Services talk to each other over Docker networking through REST APIs built with Flask.
Services expose health checks and runtime metrics, which feed continuous monitoring by the Monitoring Engine.
Structured logging across services keeps failures easy to trace.
Key Features
Proactive failure detection across distributed services.
Health checks, runtime metrics, and structured logs for every service.
Modular architecture with scalable backend components, designed to extend into incident-management workflows.
AI-Powered PDF Structure & Relevance Extractor — a two-stage pipeline that reconstructs document hierarchy and ranks sections by relevance to a given persona, built to run fully offline.
Overview
A document intelligence system that takes any PDF and recovers its true structure — headings, sections, hierarchy — using layout signals rather than relying on the file's own metadata, then ranks content by relevance to a specified persona.
How It Works
Stage 1 — Structure: PyMuPDF extracts text spans with font metadata; layout-aware heuristics use those signals to rebuild the heading hierarchy.
Stage 2 — Relevance: the recovered sections are ranked semantically with NLP against a given persona to pull out the content that matters to them.
Everything runs offline inside a Docker container, so results are reproducible on any machine.
Key Features
Two-stage pipeline in Python reconstructing document hierarchy via layout-aware heuristics and font metadata.
Semantic ranking of document sections using NLP to extract persona-relevant content.
Fully offline execution — no external API dependency at inference time.
Containerized with Docker for portability and reproducibility.
A backend system for managing users and financial records, built with role-based access at its core.
Overview
A REST API service for handling financial records across multiple user roles, with authentication, authorization, and aggregation built in from the ground up rather than bolted on.
How It Works
Users authenticate and receive a JWT, which is verified on every protected request.
Role checks (viewer, analyst, admin) decide which routes and operations each user can reach.
Records live in MongoDB; dashboard endpoints use aggregation queries instead of computing totals client-side.
Key Features
Built with Node.js, Express, and MongoDB to manage users and financial records with role-based access control.
JWT-based authentication and authorization supporting viewer, analyst, and admin roles.
REST APIs for CRUD operations and filtering of financial data.
Aggregation-based endpoints purpose-built for dashboard insights.
Smart Home Energy Management System — an AI prototype that simulates household energy usage and turns it into practical, appliance-level recommendations.
Key Features
Prompts designed to analyze household energy usage patterns.
Prompt-based workflows that simulate consumption scenarios to derive optimization strategies.
Actionable appliance-usage recommendations — up to 25% lower energy use across simulated test scenarios.