Available for new projects
Full Stack Developer
Debjit Dey
Full Stack·Cloud·Architecture·GenAI
Full-stack engineer with close to a year of production experience across two companies. Delivered frontend performance wins, backend services, and cloud deployments. Comfortable owning features end to end.
उद्धरेदात्मनात्मानं नात्मानमवसादयेत्Elevate yourself by yourself
Bhagavad Gita
— 6.5
— 6.5
01. About Me
Building elegant robust systems with precision.
As a Full Stack Engineer with nearly a year of production experience, I specialize in delivering end-to-end features that bridge sophisticated frontends with high-performance backend systems.
My work spans across Next.js performance optimization, scalable Node.js microservices, and seamless GenAI integrations. I thrive on owning complex features from architecture to deployment.
I hold a B.Tech in Computer Science Engineering from Techno India University with a CGPA of 8.31. A Google Cloud Digital Leader, I am constantly exploring the intersection of cloud scalability and intelligent automation.
श्रद्धावान् लभते ज्ञानम्One who has faith obtains knowledge
Bhagavad Gita
— 4.39
— 4.39
React / Next.jsNode.jsTypeScriptGCP / AWSGenAI / LLMs
1.5+
Years Exp.
12+
Projects
8+
Tech Stacks
8.31
CGPA
02. Experience
Career History
नियतं कुरु कर्म त्वं कर्म ज्यायो ह्यकर्मणःAction is superior to inaction
Bhagavad Gita
— 3.8
— 3.8
03. Projects
Selected Works
यतः प्रवृत्तिर्भूतानां येन सर्वमिदं ततम् । स्वकर्मणा तमभ्यर्च्य सिद्धिं विन्दति मानवः ॥By worshiping through one's own work, a person attains perfection
Bhagavad Gita
— 18.46
— 18.46
01Airspace Intelligence / ML
SkyWatch Live — Airspace Intelligence Platform
ReactDjangoCeleryRedisPostgreSQLPython
11,800+ tracked aircraft
- Built a real-time airspace surveillance platform tracking 11,862 active aircraft (10,894 airborne) on 30-second polling cycles, ingesting live ADS-B state vectors from 5+ public feeds (OpenSky, ADS-B One, FLARM) with cross-source deduplication so each aircraft appears only once on a Leaflet canvas map.
- Implemented a 3-model ML ensemble (Isolation Forest, Local Outlier Factor, MLP Autoencoder) over a 30-dimensional engineered feature space to flag trajectory anomalies asynchronously via Celery, achieving a 0.38% detection rate across active vectors.
- Indexed 85,390 airport facilities across 249 countries for dynamic origin/destination resolution; integrated SGP4 orbital propagation of CelesTrak TLE data for satellite footprint overlay, and pulled live FAA TFR and AWC SIGMET feeds to display active airspace constraints alongside flight traffic.
- Designed an event-driven backend with Django Channels, Celery workers, and Redis so ML scoring and data ingestion run asynchronously without blocking the WebSocket telemetry stream; exposed a Prometheus /metrics endpoint with 8 custom gauges, counters, and histograms feeding a Grafana observability dashboard.
02Distributed Systems / Infra
ARCE — Distributed Adaptive Traffic Control System
TypeScriptNode.jsRedisLuaDocker
Adaptive traffic control
- Implemented token bucket, sliding window, and leaky bucket algorithms with Redis-backed Lua scripts for atomic, race-condition-free enforcement.
- Built a policy engine that dynamically downgrades traffic tiers based on behavioral signals — burst patterns, route scanning, repeated fingerprints.
- Ships with an Express middleware SDK and a real-time metrics dashboard for live traffic monitoring.
03AI / EdTech Platform
GrindNode — AI Interview Prep
React 18TypeScriptPostgreSQLDockerAWS EC2Gemini API
Gemini-powered hinting
- Containerized Piston code execution runtimes with Docker on AWS EC2, enforcing per-submission CPU/memory quotas by subscription tier.
- Built a custom judge engine supporting exact match, floating-point tolerance, and tree/graph structure validation for edge cases.
- Integrated Razorpay with idempotent webhook listeners for payment events, handling prorated plan upgrades and automated invoicing.
- Wired Monaco editor to Gemini API to deliver context-aware hints based on user code, reducing average time-to-solve by ~25%.
Sub-200ms synchronization
- Architected a real-time booking system with 3 separate clients (customer, driver, admin) over a single Node.js/Socket.io backend.
- Implemented live driver location tracking with Socket.io, including robust reconnection handling to recover from dropped mobile signals.
- Secured all API routes with JWT-based RBAC, enforcing role boundaries for customers, drivers, and admins with zero data leaks.
- Built a live admin panel displaying active ride counts and driver availability, updated instantly via WebSocket events.
04. Capabilities
Technical Arsenal
Languages
TypeScriptJavaScript (ES6+)PythonLua
Frontend
React.jsNext.jsReact NativeTailwind CSSJest
Backend
Node.jsExpress.jsDjangoREST APIsWebSocketsSocket.ioCeleryGemini APIRazorpay
Databases
PostgreSQLMySQLMongoDBRedis
ML & Data
scikit-learnKerasIsolation ForestLocal Outlier FactorMLP AutoencoderLSTMFeature Engineering
DevOps & Cloud
DockerAWS (EC2, S3, CloudFront)GCP (Compute Engine)GitGitHub ActionsCI/CD
न हि ज्ञानेन सदृशं पवित्रमिह विद्यतेThere is nothing as purifying as knowledge
Bhagavad Gita
— 4.38
— 4.38
05. Contact
Let's Build Together
Currently open to new opportunities. Whether you have a project in mind or just want to say hi — I'm always happy to connect.