Skip to content
View sepehrmn's full-sized avatar
🐳
building something great.
🐳
building something great.

Block or report sepehrmn

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
sepehrmn/README.md

Sepehr Mahmoudian - AI Systems Architect | Generative Media Tinkerer

Hi there 👋 I'm Sepehr.

Software engineer specializing in artificial intelligence, machine learning, and production-grade multi-agent systems. Experienced in full-stack development with TypeScript/React, Python/PyTorch, multi-cloud infrastructure (Cloudflare, AWS, GCP, Azure), and edge computing architectures. Expert in designing and implementing strategic multi-cloud architectures that optimize performance, minimize vendor lock-in, and maintain lean operations with cost efficiency and minimal latency overhead.

I'm currently building Torus Automations, which is a first-of-its-kind Web3 product with AI-powered content generation tools and sovereign AI for data and content ownership and privacy.

Highly experienced with AI-powered development tools as an integral part of my daily workflow, achieving productivity and output quality that far exceeds what any single developer could accomplish without these tools, all while maintaining exceptional code quality and architectural standards. I continuously evaluate and integrate cutting-edge AI development tools, typically adopting new tools into my workflow on a monthly basis to stay at the forefront of developer productivity.

Socials

The best way to reach me is via email or on X. Check out my work at Torus Automations

LinkedIn   Substack   Google Scholar   X (Twitter)    Hugging Face

Expertise

I mainly use Python and its ecosystem for AI/ML, but nowadays I am using TypeScript more and more.

Over the years, I've used many machine learning models, including reinforcement learning, different recurrent neural networks, Bayesian modeling, and various methods for clustering, classification, regression, and dimensionality reduction.

Skills & Technologies

🤖 AI/ML Python
Python
PyTorch
PyTorch
Numpy
Numpy
Pandas
Pandas
SciPy
SciPy
Pydantic
Pydantic
⚙️ DevOps/MLOps Docker
Docker
Kubernetes
Kubernetes
Terraform
Terraform
git
Git
☁️ Cloud Platforms Cloudflare
Cloudflare
GCP
GCP
AWS
AWS
Azure
Azure
🔧 Backend & Infrastructure C
C
C++
C++
Python
Python
FastAPI
FastAPI
Drizzle ORM
Drizzle
Better Auth
Better Auth
Neon PostgreSQL
PostgreSQL
🎨 Frontend & Web JavaScript
JavaScript
TypeScript
TypeScript
React
React
Vite
Vite
TanStack
TanStack
Zustand
Zustand
SparkJS
SparkJS
📱 Mobile React Native
React Native
Expo
Expo
🖥️ Compute Providers FalAI
FalAI
Runware
Runware
ElevenLabs
ElevenLabs
📊 Misc Matplotlib
Matplotlib
Three.js
Three.js
LaTeX
LaTeX

My most important skills are adapting quickly, picking up new tech fast, and executing relentlessly with perseverance. I think these will be more important than ever with how fast technology will be advancing.

Life philosophy

In a World that is constantly changing, not taking any risk is the biggest risk of all.

Pinned Loading

  1. torus-automations/groupweave torus-automations/groupweave Public

    Open source infrastructure for user-owned AI, co-creation and co-immersion of genAI content

    TypeScript 5 4

  2. brojapid-activationfunctions brojapid-activationfunctions Public

    Partial Information Decomposition (PID) analysis of activation functions for PIDeepnets (PID deep neural networks)

    Python 3

  3. mahmoudian-2020-rescience mahmoudian-2020-rescience Public

    Three-way information-theoretic analysis of transfer functions - Sepehr Mahmoudian's replication of Smyth, Phillips, Kay 1996

    Python 3

  4. unsloth unsloth Public

    Forked from unslothai/unsloth

    Finetune Llama 3.3, Mistral, Phi, Qwen 2.5 & Gemma LLMs 2-5x faster with 80% less memory

    Python