Hey, Hesam here
I like tinkering with AI models, agents, harnesses, model architecture, practically any kind of matrix multiplication performed by rocks.
Since you arrivedone Vera Rubin GPU000.000 quadrillion FLOPs
Designed by Opus 5.5
Experiences
- Jan 2026 – PresentIndependent Open-Source AI ProjectsOpen-Source AI Engineer, Remote
- Grazing in the field of decentralized open AI: fine-tuning and releasing open models, and building reinforcement learning environments for LLM agents.
- OctoGent, a dashboard for orchestrating Claude Code.
- Tock, a desk pet and focus timer for the M5Stack FIRE.
- Jun 2025 – Dec 2025CAMEL AIAI Engineer (Graduate Internship), Remote (London, UK)
I helped maintain CAMEL’s open-source multi-agent library, reviewing what the community sent in, and built an episodic memory for its agents that did better on GAIA while spending fewer tokens.
- Mar 2024 – May 2025Towards Data ScienceAI Technical Writer, Remote
I wrote articles on LLMs, agents, NLP and reinforcement fine-tuning: how they work, and when they’re worth using.
- Mar 2024 – May 2025DataCampAI Technical Writer, Remote (New York, USA)
I wrote hands-on tutorials on LLMs, agents and Hugging Face tools, one working step at a time.
- Jun 2021 – Jul 2023PFKMachine Learning Engineer
I built real-time computer vision for sorting agricultural produce and reading licence plates, with CNNs, vision transformers and YOLO, tuned with TensorRT to run on the machines themselves.
I checked the chapters for technical accuracy before the book went to print.
Projects
- Star1.4k
octogentPublic
A thin orchestration dashboard over Claude Code. You need tentacles.
TypeScript1,419 stars244 forks
- Star1
- Star2.4k
ml-retreatPublic
A machine learning journal for intermediate to advanced topics.
Jupyter Notebook2,376 stars251 forks
- Star31.7k
awesome-openclaw-usecasesPublic
A community collection of OpenClaw use cases for making life easier.
- 01"bro, I want to become an AI Engineer this year" - here's what NOT to do.there's a thousand things to learn, you can't find a solid roadmap, everyone seems to be doing stuff and FOMO gets you. the pain is real, because you're doing it wrong.
- 02the AI agent paradigm has shifted. here's why. (no hype, I promise)if that last two weeks left you confused about ClawdBot, Moltbook, and what the fuck is going on in AI, you're not alone. let's break down what really matters.
- 03everyone talks about Clawdbot, but here's how it worksI took a look inside Clawdbot (aka Moltbot) architecture and how it handles agent executions, tool use, browser, etc. there are many lessons to learn for AI engineers.
- 04you need to become a polymath (or get replaced by machines)if you're anything like me, you've been told your whole life that success requires focus. pick a thing, learn it for years, do it for the rest of your life. that's crap.
- 05Learning 2.0: the shift most people will missthe way you were taught to learn is optimized for teachers, not you. and it's costing you years of wasted effort. AI removed that barrier but you're still acting like it's there.
- 06Why you're still slow even with AIMost of our old habits are now optimizing for the incorrect thing.
- 07a letter to my fellow CS studentsI know you're confused and disappointed.
Collaboration
Building something in one of these?
A company, a lab, an open-source project: if it’s genuinely exciting, I’m open to joining the team or collaborating on it. Tell me about it.
My interests
- AI safety & alignment
- Inference engineering
- Machine learning
- Mechanistic interpretability
- AI agents
- Agentic harnesses
Show me what you’re building
Send it over and I’ll try it myself. If it’s useful to the people who follow me, mostly programmers and AI-native builders, we’ll get more eyes on it.