• 🌐 Portfolio: https://ateet.masxai.com
• 💼 Projects: https://ateet.masxai.com/projects
• 📝 Blog: https://ateet.masxai.com/blog
• 🔗 LinkedIn: https://www.linkedin.com/in/ateet-vatan-bahmani
• 📧 Email: ab@masxai.com
• 🧪 Galileo Live Demo: https://galileo.masxai.com/

I’m building MASX AI as a modular, production-grade platform for real-time signal ingestion, agentic reasoning, and forecasting.

MASX AI Modules

1. MASX Forecast — Global Predictive Analytics + Flashpoint Detection + Feed Generation. I have built an autonomous system that ingests raw telemetry and intelligence signals to analyze global datasets, predict future flashpoints, and automatically produce broadcast-quality intelligence briefings to keep decision-makers ahead of the curve.

   https://forecast.masxai.com/ (closed source)

2. MASX GeoSignal — high-throughput ingestion + enrichment layer + Global Flashpoint content Generation

   https://geosignal.masxai.com/ (closed source)

Multi-model agentic debate evaluation platform implementing the “Galileo Test” style evaluation harness.
It evaluates LLMs via adversarial deliberation (Orthodox / Heretic / Skeptic / Judge), with deterministic + ML hybrid scoring and a full audit trail.

• 🔥 Live Demo: https://galileo.masxai.com/
• 📦 Repo/Docs: https://github.com/AteetVatan/AIGalileoArena

Voice-enabled personal AI clone platform designed to answer questions about Ateet’s background, projects, and expertise through conversational AI. It enables interactive text + voice Q&A, with a structured knowledge layer, retrieval-backed responses, and a portfolio-style AI experience.

• 🔥 Live Demo: https://ateetclone.masxai.com/
• 📦 Repo/Docs: https://github.com/AteetVatan/echo-ai

• Agent SDK patterns, LangChain, LangGraph, AutoGen
• RAG (retrieval + grounding), tool/function calling, structured outputs
• Hugging Face ecosystem, PyTorch, TensorFlow
• Workflow automation: n8n

• Python, FastAPI, Flask, Pydantic
• C# / .NET (platform + API work)
• Vector search & embeddings: Chroma, FAISS (and scalable retrieval patterns)
• Datastores: PostgreSQL / Supabase, SQL Server
• Real-time + async pipelines, orchestration-first architecture

• LangSmith, PromptLayer, TruLens
• Prompt versioning, deterministic runs, cost tracking
• CI/CD for agent systems, regression-style eval gates, audit-friendly execution

• Angular (v15–17), TypeScript, Tailwind
• Chat UIs, agent dashboards, evaluation panels

• AWS: Lambda, ECS, S3, API Gateway, IAM
• Docker, GitHub Actions, CI/CD automation

• Cybersecurity mindset + data integrity
• Compliance-driven development, stakeholder alignment, change management

• 🎙️ EchoAI — Real-Time Voice-Driven Agentic Intelligence Voice-first AI clone with real-time STT → RAG memory retrieval → LLM reasoning → TTS, built for autonomous conversations with persistent, grounded knowledge. Highlights: WebSocket audio streaming, Faster-Whisper (local) + Whisper fallback, Edge-TTS, ChromaDB semantic memory, Supabase + SQLite caching, DeepSeek (primary) + Mistral fallback, LangChain RAG. Repo: https://github.com/AteetVatan/echo-ai

• 🚀 ApplyBots — AI-Powered Job Application Platform Agentic platform to discover jobs, generate tailored applications, and submit safely with Truth-Lock (resume-verified, anti-hallucination) + human-in-the-loop approval. Includes ATS automation (Greenhouse/Lever) with audit logs/screenshots, campaigns, and analytics.
  Repo: https://github.com/AteetVatan/ApplyBots

• 🤖 medAI — Clinical Speech & Documentation Platform Real-time clinical intake + documentation backend for German-speaking clinics: speech-to-text, medical entity recognition, LLM summarization, translation, and PDF report generation — orchestrated as an agentic pipeline.
  Repo: https://github.com/AteetVatan/medai.git

• 🎬 ShotGraph — AI Cinematic Video Generation Pipeline Multi-agent pipeline that converts long-form stories into cinematic videos using LLM scene breakdown + shot planning, diffusion video generation, TTS narration, music generation, and MoviePy/FFmpeg composition.
  Repo: https://github.com/AteetVatan/ShotGraph

• 🚀 AgenticMatch-3H
  3-hour Agentic AI prototype that lets users upload an image and matches it to partner brands based on visual style and mood using CLIP embeddings and FAISS similarity search.
  Tech Stack: Python, CLIP, FASIS, FastAPI https://huggingface.co/spaces/AteetVatan/agenticmatch

• 🚀 Movie Management System
  Full-stack movie database application with CRUD with MVC pattern and API integrations.
  Tech Stack: Python, Flask, SQLite, Jinja2, Bulma CSS https://movie-app-kufy.onrender.com

• 🚀 Nexora Portfolio Application
  portfolio website built with React, TypeScript, and Tailwind CSS.
  Tech Stack: React, TypeScript, Tailwind CSS, Vite, Supabase, Postgres https://ateetai.vercel.app/

• 🚀 Masterblog API MVC-inspired, Repository Pattern, Service Layer, modular API for dynamic blog content management, following clean architecture and REST principles. Tech Stack: Python, Flask, SQLAlchemy, SQLite, Jinja2,

• 🚀 Drone Delivery Planner
  Real-time drone delivery optimization with dynamic geospatial intelligence.
  Tech Stack: c#, AWS , Map APIs, Optimization Algorithms

• 🚀 MASTER BLOG Flask MVVM CRUD Application
  Flask-based web application following the MVVM (Model-View-ViewModel) pattern with CRUD. Tech Stack: Python, Flask, Jinja2

• In the repository you can find more projects.---

MASX AI is my long-term build: a Strategic Intelligence OS — the Palantir of the future, rebuilt for the AI era: open, agentic, measurable, and evidence-first.

To help guide humanity through complexity — by turning global signals into decision-grade intelligence that is grounded, transparent, and accountable (not black-box “trust me bro” outputs).

• Always-on Global Signal Grid: ingest + normalize multilingual OSINT (RSS / GDELT-style feeds / structured sources).
• Entity + Event Intelligence Layer: dedupe, link, timeline, and provenance for who/what/where/when.
• Agentic Reasoning + Doctrine Engine: multiple strategic agents debate, validate, and simulate outcomes (incl. counterfactuals).
• Forecasting (measurable, not vibes): predictions with confidence, time horizons, assumptions, and “what would change my mind” triggers.
• Evidence-first Outputs: every insight links back to sources, extracted claims, and reasoning steps — audit trail by default.
• Eval-Gated Quality Control: continuous regression eval + scoring so quality improves systematically over time.
• Strategy Cockpit UI: dashboards for hotspots, timelines, confidence, maps, and longitudinal accuracy tracking.

• 2026: Stable daily system publishing evidence-backed hotspots + forecasts.
• 2027: Multi-source fusion + stronger entity/event graph + scalable orchestration.
• 2028: Forecast calibration + feedback loops (accuracy scoring, auto-refinement).
• 2029: Multimodal signals + richer simulations and scenario planning workflows.
• 2030: Modular platform: plug-in agents, plug-in doctrine packs, deployable for real teams.

North Star: help societies make better calls — faster, calmer, and based on evidence.