"Building production systems under pressure. Not just coursework."
> whoami
404Avinash — Avinash Jha
> role
2nd Year CSE @ ABES Engineering College, Ghaziabad
> focus
Cloud Infrastructure · CI/CD Pipelines · Real-Time Systems · DevOps
> expertise
Docker · AWS EC2 · GitHub Actions · Terraform · Redis · WebSockets
> currently_building
Production systems across multiple national-level hackathons
Real-time fraud detection · Railway safety platforms · AI middleware compression
> location
~/building/infrastructure_that_scales| Achievement | Event | Role | Result |
|---|---|---|---|
| 🏅 National Winner | Decode SIH OsCode, IIIT Delhi (2025) | Team Lead & DevOps | ₹30,000 Selected from 15,000+ registrations, 3,500+ teams |
| 🥇 1st Place | Eureka IIT Bombay NEC 2026 Internals | Team Lead | moved to zonals round |
| 🥈 2nd Place | Claw & Shield, IIT Delhi TRYST '26 | Team Lead & System Architect | ₹12,000 |
| 🥉 2nd Runner Up | Devcation, IIT Delhi × IGDTUW (2026) | Team Lead & DevOps | ₹8,000 |
| 🏅 National Winner | CSIR-NAL Aerofest 3.0 (2023) | Team Lead | High-endurance HAPS UAV prototype |
| 🏅 National Winner | CSIR-NAL Aerofest 2.0 (2022) | Team Lead | Aerodynamic electric aircraft prototype |
🚂 DRISHTI — Production Railway Safety Platform (Devcation IIT Delhi × IGDTUW '26)
Live Deployment: 44.216.1.62 | Repo: github.com/404Avinash/drishti
2nd Runner Up | Team Lead
Built a full-stack production system handling real railway safety under competition pressure.
- Containerization: Dockerized full stack with Docker Compose API (FastAPI), frontend, Redis, and telemetry service — isolated containers with shared networking
- CI/CD Pipeline: GitHub Actions for automated linting, pytest runs, Docker image builds, zero-downtime deployment to AWS EC2 on every push to main
- Infrastructure as Code: Provisioned cloud infrastructure using Terraform on AWS EC2; managed IAM, security groups, and SSH access for team deployment
- Real-Time Telemetry: Redis pub/sub and WebSockets pipeline — live alert delivery from backend risk engine to frontend dashboard
- Multi-Model Risk Layer: Bayesian Network, Isolation Forest, DBSCAN, and Causal DAG designed to eliminate single-model blind spots in anomaly detection
Tech Stack: Docker · Docker Compose · FastAPI · Redis · WebSockets · GitHub Actions · Terraform · AWS EC2 · Python
💳 SAMPATI — Real-Time UPI Fraud Detection System (Decode SIH OsCode '26)
Live Deployment: http://52.66.244.253:8000/ | Repo: github.com/404Avinash/SAMPATI_V2
🏅 National Winner | Team Lead | Coordinated 3,500+ competing teams
Designed and deployed a fraud prevention backend processing live UPI transactions using pure graph topology — no ML model dependency, deterministic and low-latency.
- Architecture Decisions: Led system design, API design, and backend deployment at national scale
- Graph-Based Detection: Pure topology-driven fraud detection eliminating single-model blind spots
- Deterministic Processing: Low-latency, dependency-free fraud scoring from transaction graphs
- Team Coordination: Coordinated across 3,500+ competing teams as one of the national winners from 15,000+ registrations
Tech Stack: FastAPI · Python · Graph Algorithms · API Design · Backend Deployment
🧠 MONDAY — AI Context Compression Middleware (Eureka IIT Bombay '26)
Eureka IIT Bombay 1st Place | Team Lead | In Research Phase
Built middleware to compress LLM context using DeepSeek vision tokens and OCR — reduces token consumption and API costs without losing semantic context.
- Vision Token Compression: DeepSeek vision tokens + OCR for intelligent context reduction
- Cost Optimization: Measurable reduction in token consumption and API costs
- Semantic Preservation: Maintains semantic context throughout compression pipeline
- End-to-End Delivery: System design, API integration, and deployment
Tech Stack: Python · FastAPI · Vision APIs · OCR · LLM Integration
| Domain | Stack |
|---|---|
| Cloud & Infrastructure | AWS EC2 · Terraform (IaC) · Linux (Ubuntu) · SSH · VPC basics |
| Containers & Orchestration | Docker · Docker Compose · Multi-service stack management (API, frontend, Redis, telemetry) |
| CI/CD | GitHub Actions · Automated linting · pytest · Container builds · EC2 deployment pipelines |
| Monitoring & Messaging | Redis pub/sub · WebSockets · Real-time telemetry · Alert delivery |
| Testing & Debugging | pytest · Postman (API testing) · Unit testing · Integration testing · Software Testing Fundamentals |
| Languages & Scripting | Python (primary) · Bash · C++ (basic) |
| Version Control | Git · GitHub branching · PRs · Multi-contributor workflows |
Production systems built under competition pressure. Real infrastructure decisions. Real deployments.