I see patterns in the noise before the noise knows it's a pattern.
I'm from Kerala. I live in Dublin. I build things because I can't not build them.
On November 11, 2024, I wanted to build something like PyTorch — but for agents. Not a wrapper. Not another LangChain. A ground-up rethink of what agentic infrastructure should actually be.
109 days later I had Lár, Lár-JEPA, DMN, and a compliance architecture mapped to the EU AI Act. I didn't plan most of it. The ideas emerged as I followed the logic of the problem wherever it led.
That's how I build everything.
Lár — The Glass-Box Agent Engine
The PyTorch for Agents.
Most agent frameworks are black boxes. When they fail in production you get a 100-line stack trace and no idea what happened, why, or how much it cost. I built Lár because I think that's the wrong foundation for serious AI systems.
Lár is a deterministic, define-by-run graph execution engine. Every node, every state change, every decision is logged to a forensic flight recorder. Built-in HMAC cryptographic audit trails. EU AI Act Article 12/13/14 compliant by architecture. Air-gap capable. FDA 21 CFR Part 11 ready.
Not magic. Not wrappers. Just pure, debuggable Python you actually own.
The numbers: 1% LLM + 99% code hybrid architecture. 0.08s vs 64s latency. $0.00 vs $3.60/run vs standard frameworks. LangGraph hits recursion limit at step 25. Lár ran 10,000+ steps without a single error.
Lár-JEPA — Post-LLM Orchestration
The nervous system for world models.
Every major agent framework assumes text in, text out. That assumption breaks when your model outputs a 768-dimensional tensor representing the abstract state of a physical environment.
Lár-JEPA is the deterministic execution spine for Predictive World Models — routing high-dimensional latent tensors without text prompting. The industry is building the brain. This is the nervous system.
DMN — Bicameral Memory Architecture
An organism, not a tool.
Standard agents have amnesia. DMN implements a biologically-inspired Default Mode Network — a background daemon that watches execution logs, sleeps, dreams, and consolidates short-term memory into long-term semantic understanding via ChromaDB. Solves catastrophic forgetting without retraining. Gets smarter while you're away.
Metacognition — Dynamic Self-Modifying Graphs
Agents that rewrite their own execution topology at runtime. Safely.
DynamicNode generates JSON graph specs — including BatchNodes for parallel execution — validated by a deterministic TopologyValidator. Self-modification as an auditable event, not a jailbreak risk. The AI proposes. The code decides.
BreakHis Classifier — ResNet-50 breast cancer classifier on histopathology data. 0.96 F1-score, 0.98 AUC.
MCP Forensic Toolkit — AI-enabled digital forensics via Model Context Protocol.
MCP BioForensics — Clinical trial data exploration with hybrid retrieval and natural-language querying.
The industry is building the Brain. I'm building the Nervous System.
Never use an LLM to police another LLM. Use code. An approval is not a flag. It is a cryptographic signature of a specific state. Self-modifying code is only dangerous in a black box. In a glass box it's just evolution with an audit trail.
Lár → deterministic execution
Lár-JEPA → world model orchestration
DMN → persistent bicameral memory
Metacognition → safe self-modification
Execution spine → world modelling → persistent memory → self-awareness.
A complete cognitive architecture. Built from scratch. In public. Under Apache 2.0.
MSc Data Analytics, Dublin City University. Kerala → Dublin. Future: CTO, SnathAI.
[email protected] · LinkedIn · snath.ai · docs.snath.ai
"Apna time aayega."