Backend
Engineer
I build resilient backends and AI-integrated systems — built to ship and hold up under load.
Notable companies I've worked for
Kiwi
$43M funding · 3M users
DMG
$1B valuation · $300M revenue · INC5000
About
Me
I build production-grade backends with FastAPI, Spring and Node.js — equally comfortable architecting monoliths or event-driven micro-services that stay fast under load.
Deep in the AI / LLM space — shipping MCPs, toying with Claude skills, integrating Langfuse for AI observability, and orchestrating reliable workflows with Temporal and Airflow.
Wired for performance & reliability — I instrument everything with Datadog, SigNoz, and Grafana, containerise with Docker, orchestrate with Kubernetes, and ship to AWS with CI/CD baked in.
Stack
& Tools
Python
Primary Language
FastAPI
API Framework
Spring
Java Framework
Redis
In-Memory Store
TypeScript
Typed JavaScript
MongoDB
NoSQL Database
AWS
Cloud Platform
Claude
AI / LLM
Docker
Containerisation
Kubernetes
Container Orchestration
Datadog
Observability
Kafka
Event Streaming
Work
Experience
Software Developer Engineer-1
- ◦Implemented job invite and photo upload pipelines using Temporal, orchestrating media ingestion, validation, and processing with reliable retries, idempotency, and full workflow traceability via Temporal UI.
- ◦Designed an analytics pipeline that pulls mobile app performance data from Datadog APIs, processes 240+ hourly data points across multiple key widgets, and delivers insights to Snowflake and MongoDB.
- ◦Integrated Langfuse to monitor LLM activity, capturing prompt flows, model latency, and errors to improve visibility into AI performance and optimize reliability and cost.
- ◦Improved data collection speed from 1.5hrs to 20 mins adding connection pooling, intelligent retries, and pagination — achieving 100% accuracy in production monitoring.
Sept 2025 — Present
Backend Engineer Intern
- ◦Worked on 4 Spring Boot microservices (card-provisioning, vault, Yes Bank adapter, polling) refactored into onboarding core following various design principles.
- ◦Built an Airflow DAG that automatically retries failed LOS pushes for YES Bank users by pulling recent KYC-approved applications, validating backend sync issues, and invoking the LOS push API for recovery.
- ◦Reduced Signoz logging cost by 40% by caching multiple states of VKYC and optimized log streaming by setting various severity levels for the logs.
- ◦Integrated Meta CAPI to restore iOS conversion tracking, increasing attribution coverage by 28% after iOS 14.5 privacy updates.
Jan 2025 — Jun 2025
NeuralSpace.ai
Full-Stack Engineer Intern
Full-stack development on the LocAI platform — AI subtitle generation from video. Cut front-end build time by 85% through dependency pruning, lazy loading, and caching.
Jun 2024 — Aug 2024
Hyperspec.ai
Developer Intern
Built high-performance backend services for the Hyperlite sensor pipeline and delivered a real-time web HMI for device control and data visualisation.
Sept 2023 — Apr 2024
I‑Hub DivyaSampark
UG Intern
Built a vehicle speed detection prototype and ANPR-based number plate recognition system for centralized traffic monitoring.
Oct 2022 — Aug 2023
Recent
Projects
SnapTrace
Cybersecurity tool that detects threats and quarantines compromised servers by analyzing real-time system and network logs. Fine-tuned Mistral-8B on the CIDDS dataset.
ClearDrive
Shows the safest roads during poor visibility conditions to prevent accidents using real-time weather and maps data.
Live Code
Collaborative real-time code editor built for hackathons — pair programming in the browser with live sync.
SmartWatch Stressor
Real-time stress level display from ML-processed health vitals streamed from an IoT wearable.
SnapTrace
Cybersecurity tool that detects threats and quarantines compromised servers by analyzing real-time system and network logs. Fine-tuned Mistral-8B on the CIDDS dataset.
ClearDrive
Shows the safest roads during poor visibility conditions to prevent accidents using real-time weather and maps data.
Live Code
Collaborative real-time code editor built for hackathons — pair programming in the browser with live sync.
SmartWatch Stressor
Real-time stress level display from ML-processed health vitals streamed from an IoT wearable.