oss @ AutoGpt,Biome,Aisoc
intern @cwistudio
https://github.com/mangod12

Anshaj Kumar • Peerlist

Production ERP for steel fabrication, deployed and running daily operations at Kumar Brothers (Bokaro). Scope: - 172 API endpoints across 3 versioned APIs (v1 legacy CRUD, v2 lot-level inventory, v3 drawing-based production tracking) - Drawing-based Kanban production tracking with per-component stage management - Dual inventory system: simple counts (v1) + lot-level traceability with immutable audit trail (v2) - Goods Receipt Note (GRN) and Dispatch workflows with weighbridge integration - Auto material deduction at fabrication stage with double-deduction prevention - 6-role RBAC (Boss, Supervisor, Store Keeper, QA Inspector, Dispatch Operator, View-only) - Excel import with 90+ column alias matching, auto-linking profiles to raw materials - Scrap management with recovery and loss analytics Infrastructure: - FastAPI 0.128 + SQLAlchemy 2.0 + PostgreSQL on Google Cloud SQL - Deployed to Google Cloud Run via GitHub Actions + Cloud Build (Workload Identity Federation) - 78 end-to-end tests running against live deployment - JWT auth with python-jose + bcrypt Live: kbsteel-backend-498310931350.asia-south1.run.app GitHub: github.com/mangod12/industryERP

IndustryERP — Steel Fabrication Management Sy

Developed a full-stack vulnerability intelligence SaaS platform that automates CVE aggregation, asset correlation, and alert generation. Key Contributions: • Designed multi-source scraping engine (NVD, MSRC, Cisco, Red Hat) • Built async FastAPI backend with SQLAlchemy 2.0 • Implemented JWT-based auth + GitHub OAuth • Developed correlation engine to match vulnerabilities with user assets • Dockerized deployment with PostgreSQL + Nginx • Implemented email alerting via Mailgun Tech Stack: Python, FastAPI, PostgreSQL, SQLAlchemy, Docker, APScheduler, JWT, GitHub OAuth GitHub Link: https://github.com/mangod12/cybersecuritysaas

CyberSec Alert SaaS

Multi-agent AI system for crisis logistics coordination — personal project exploring agentic AI architectures. - Multi-agent control plane with Gemini 2.5 Flash as the reasoning model - MCP (Model Context Protocol) for standardized tool execution across agents - PostgreSQL + pgvector for RAG-grounded agent memory (situation reports, resource catalogs) - FastAPI orchestration layer, deployed on GCP Cloud Run Stack: Python, FastAPI, Gemini 2.5 Flash, MCP, PostgreSQL, pgvector, Docker, Cloud Run GitHub: github.com/mangod12/TaskForge

TaskForge

• Built Arduino-based sensor integration systems • Worked with drone telemetry data and hardware communication • Developed backend dashboards for real-time monitoring • Experience with high-voltage and industrial electrical environments

IoT & Drone Systems (Independent Projects)

Open-source OT/ICS cybersecurity platform. Solo developer and maintainer. 26 stars on GitHub, MIT licensed. Positions as the affordable, self-service alternative to enterprise OT vulnerability platforms (Claroty, Dragos, Nozomi) — same-day deployment instead of weeks, $499–$4,999/mo instead of $300K–$800K/yr. 10 phases shipped across security, frontend, multi-tenancy, AI remediation, compliance, SBOM, network topology, billing, integrations, and observability: - Multi-source CVE aggregation (NVD, CISA KEV, ICS-CERT, Cisco PSIRT, Microsoft MSRC, Red Hat) - AI-powered remediation engine with 5 OT-aware rules (patch vs. compensating control based on Purdue zone) - EPSS exploit probability scoring + CISA KEV active exploitation flagging - Compliance-as-code: IEC 62443-3-3 (10 controls) + NIST CSF 2.0 (11 controls), continuous automated evidence collection - SBOM ingestion (CycloneDX + SPDX JSON) with component-level vulnerability cross-reference - Passive OT asset discovery via Zeek, Suricata, SNMP, Shodan - Industrial protocol detection (Modbus, DNP3, PROFINET, EtherNet/IP, OPC-UA, HART) with Purdue model zone classification - SIEM/SOAR integrations: Splunk HEC, Microsoft Sentinel, ServiceNow, PagerDuty - Stripe billing with 4-tier subscription gating - TOTP MFA, structured logging, request metrics middleware Stack: Python 3.11, FastAPI, SQLAlchemy 2.0, PostgreSQL (Cloud SQL), React 18 + TypeScript + Vite + Tailwind v4, Zustand, Docker (multi-stage), Google Cloud Run, Cloud Build, APScheduler, 166 pytest tests Live: cybersec-saas-ebqzvaqu6a-uc.a.run.app/app/ GitHub: github.com/mangod12/OneAlert

OneAlert — CyberSec Alert SaaS

Real-time drone telemetry dashboard for MAVLink-compatible aircraft (ArduPilot, PX4) — personal project exploring high-frequency streaming UIs. - 20Hz (50ms) telemetry polling with altitude history, attitude indicators (roll/pitch/yaw), battery monitoring, GPS, flight mode - MAVLink adapter for USB/serial/TCP/UDP drone connections - DJI SDK adapter scaffold for extensibility - Glass-morphism dark-theme UI with Recharts + Framer Motion - FastAPI backend with built-in drone simulator for testing without hardware Stack: Python 3.11, FastAPI, PyMAVLink, React 18, Vite, Tailwind CSS, Recharts GitHub: github.com/mangod12/skydash

SkyDash

AI-powered IVR platform for emergency dispatch and crisis management. Originally led development in a 4-person team; independently extended with Gemini integration, model retraining, CI/CD pipeline, and production deployment. My contributions: - Trained custom DistilBERT intent classifier on 50k+ annotated emergency transcripts — 94.2% accuracy, 8 classes, exported to ONNX for 15ms CPU inference - Built CNN emotion detector on RAVDESS audio data — 87.3% accuracy, continuous panic score output - Designed hybrid severity scoring engine (panic 40% + keywords 30% + incident priority 20% + urgency 10%) - Built multi-tenant FastAPI backend with row-level data isolation, WebSocket dashboard, and Celery background processing - Deployed to GCP Cloud Run with GitHub Actions CI and Prometheus observability Stack: Python, FastAPI, Whisper, DistilBERT, ONNX Runtime, Celery, Redis, PostgreSQL, Docker, GCP Live demo: redline-ai-359883234654.us-central1.run.app/demo GitHub: github.com/mangod12/redline-AI

Anshaj Kumar

oss @ AutoGpt,Biome,Aisoc intern @cwistudio https://github.com/mangod12