Skip to content
View MajidAsghariTabrizi's full-sized avatar

Block or report MajidAsghariTabrizi

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.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

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

Report abuse

Majid Asghari Tabrizi — Product strategy. AI infrastructure. Systems engineering. From intent to evidence to execution.

Portfolio   /   Quantiviq   /   LinkedIn   /   Let's talk

I build systems that can explain their decisions.

I'm Majid Asghari Tabrizi, a product strategist and AI systems builder based in Iran, and the founder of Quantiviq. I connect product judgment with the infrastructure behind it: what an agent knows, which model it calls, what it can do, and how it recovers.

My work follows one thread: make evidence, uncertainty, and failure part of the design.

Selected systems

01 / Bandpass — context compilation: rank, verify, resolve, compress.

Give agents better grounds. Compiles memory, files, state, and evidence into a compact, inspectable Context Capsule over MCP.

TypeScript · MCP · Context engineering

Try the interactive lab ↗ · Read the architecture

02 / Credence — evidence-backed memory: record, correct, resume.

The process ends. The mission continues. Evidence-linked agent memory, explicit unknowns, immutable belief history, and missions that resume from durable state.

TypeScript · Agent memory · Durable execution

Run the kill-and-resume demo ↗

03 / Free Best Router — discover, rank, route, recover.

One endpoint. A changing model landscape. Discovers and ranks free AI models, checks their health, and routes requests with bounded failover and cooldowns.

JavaScript · Model routing · Self-hosted

Get started ↗

04 / Phoenix — verify state, price uncertainty, gate execution.

Execution has to earn permission. Arbitrum financial infrastructure with independent state verification, conservative economics, and fail-closed transaction gates.

Rust · Arbitrum · Financial infrastructure

Explore the system ↗

Research with an honest answer

The Gambler / Evidence-driven trading research. A decision to do nothing can be the right result.

The Gambler is my Python research lab for causal prediction, realistic execution costs, paper trading, and governed learning. Its published findings report that none of the evaluated signals cleared the execution-cost hurdle. The negative result is part of the work.

Findings and evidence ↗ · Reproducible cost fixture

How I build

Start with the decision. Define the user, constraints, and economics before the architecture.
Keep the evidence. Preserve provenance, corrections, and unknowns.
Design the recovery. Bound retries, make authority explicit, and keep state durable.
Make the work inspectable. Ship runnable demos, clear limitations, and reproducible checks.

Also building Universal Engineering Agent: a vendor-independent reference kernel for inspecting, planning, verifying, and recovering engineering work.


Working on AI-native organizations or agent infrastructure?
Explore Quantiviq · See my engineering work · Get in touch

Popular repositories Loading

  1. free-best-router free-best-router Public

    Open-source intelligent router for automatically discovering, ranking, health-checking, and routing requests to the best available free AI models.

    JavaScript 5

  2. anti-gravity-phoenix-v4 anti-gravity-phoenix-v4 Public

    Fail-closed, AI-assisted financial infrastructure for Arbitrum: Aave V3 liquidations, Atlas auctions, DEX arbitrage research. Dual-provider agreement, conservative economic gate, submission-unknown…

    Rust 2

  3. credence credence Public

    Evidence-first agent memory and durable mission runtime — UNKNOWN states, immutable belief history, governed learning, and missions that survive kill -9. Zero-dependency TypeScript kernel.

    TypeScript 1

  4. bandpass bandpass Public

    Context compiler for AI agents — build compact, evidence-aware Context Capsules from memory, state, files and tools over MCP. Compile context, don't concatenate it.

    TypeScript 1

  5. felotin-ecommerce felotin-ecommerce Public

    Product manager and web developer

    PHP

  6. smart-trader smart-trader Public

    Python