FAQ — Straight Answers

I'm an LLM & Agentic AI Engineer. I build the full scaffolding around language models so they don't just generate text — they perform tasks: agent orchestration, tool use, memory across turns, planning, and the evaluation loops that verify the work actually happened. Today that's my full-time craft; the same instinct that earlier made me ship running ML systems instead of notebooks.

An AI swarm is a coordinated system of multiple independent agents that share goals, decompose tasks, and negotiate solutions together. Unlike a single agent — one loop with one bottleneck and one point of failure — a swarm gets resilience and scale from redundancy and division of labor: parallel exploration, no single point of failure, and answers that emerge from collective reasoning rather than one model's guess.

My working thesis is that intelligence emerges from structured interaction among many agents, not from a single bigger model.

Yes — that's the main way to work with me. I'm open to collaborations with people and companies: LLM systems, agent orchestration, tool use, memory, and multi-agent swarms. Reach me via the contact page — if it needs a verifier, an action API, or a coordination layer, it's probably a good fit. You can also collaborate with AI Sensile, the company I co-founded.

AI Sensile (aisensile.com) is the company I co-founded. It builds the AI operating layer for modern business: an AI-powered CRM, a growth engine that generates high-intent content, and an AI media suite — plus Sendwise Live, a WhatsApp API platform. The philosophy: AI is a partner, not a replacement.

My five publications are listed on the publications page and indexed on Google Scholar — all published under my former name Behzad Aslani Avilaq. They span decision science, optimisation, biomedical AI, and molecular dynamics, including the enzalutamide drug-resistance work in the Biophysical Journal.

So I can see inside every model. Writing Ridge, Lasso, ElasticNet, gradient descent versus SGD, and a Bayes classifier line by line is the best way to understand why they work — and it gives me the confidence to trust a model only when I can see inside it. That discipline grew into a public GitHub library that runs from-scratch ML through federated learning and on toward agents — implementation depth as the credibility anchor.

Brain-connectivity super-resolution: building models that recover dense, high-resolution brain-wiring graphs from coarse, low-resolution connectomes. I released the connectivity matrices as a public CC0 Kaggle dataset (289 downloads). It is also the literal seed of my thesis — studying intelligence from neurons to multi-agent systems.

Yes. The writing page holds three position essays — Why I Bet on Swarms, What Connectomics Taught Me About Multi-Agent Systems, and Evaluation Is the Bottleneck, Not the Model. They're opinion pieces, not research claims: my arguments, the evidence, and the engineering discipline behind betting on swarms.

Usually within 2–3 days. I keep email and contact messages to a short queue so nothing useful gets lost, and I reply to every serious note personally.

Vite + Three.js with vanilla JavaScript — a dark glassmorphism theme (Space Grotesk + JetBrains Mono, cyan/violet on #05060a). No frameworks, no tracking. Full details and image attributions are on the credits page.