May 2026
Trisynapse Memory
Memory layer for AI Agents
A local-first, high-performance memory architecture and engine designed for AI agents, multi-agent workflows, and local LLM research systems.
Telugu ❤️
Telugu ❤️ · English · Hindi
I'm Shanmukha Sainath, AI Engineer at KLA Corporation. I specialize in building algorithms for high-precision optical metrology, and I love coding models from scratch to understand them under the hood. Here, I document my experiments, research papers I read, and the systems I build across NLP, Computer Vision, RL and Generative AI.
Chennai, India, --:--Open to collaborations
Visited
Active Projects
Apr 2026
Trisynapse
Desktop AI Research Workspace
A local-first, privacy-focused desktop AI research assistant. Bring your own keys to compile sources into local wikis, explore concept graphs, perform semantic search, utilize an infinite whiteboard canvas, and monitor topics offline.
Sep 2026
AresSim
Mars Survival Simulation & RL Environment
A deterministic 32×32 Mars survival grid for training RL algorithms and LLM agents: a rover explores terrain, extracts ice and ore, and keeps a 5×5 colony pad alive under weather, battery drain, and seed-reproducible worlds. Playable in the browser and trainable via Gymnasium, PettingZoo, and RLlib.
Experience
Projects
Browse by area, year, or type.
Sep 2026 · Project
AresSim
Mars Survival Simulation & RL Environment
A deterministic 32×32 Mars survival grid for training RL algorithms and LLM agents: a rover explores terrain, extracts ice and ore, and keeps a 5×5 colony pad alive under weather, battery drain, and seed-reproducible worlds. Playable in the browser and trainable via Gymnasium, PettingZoo, and RLlib.
Jun 2026 · Vibecoding
tensor trace
Neural Networks Forward Pass Visualizer
An interactive, scroll-driven educational walkthrough of a single neural networks forward pass, following a token from PyTorch framework operations down to PTX assembly and GPU silicon rooflines.
Jun 2026 · Vibecoding
Memory for AI Agents
Interactive Survey & Taxonomy Explorer
A comprehensive interactive encyclopedia and survey mapping modern memory architectures for AI agents. Visualizes short-term, long-term, and multi-stage memory taxonomies, benchmarks, and memory engines with rich animations.
May 2026 · Project
Trisynapse Memory
Memory layer for AI Agents
A local-first, high-performance memory architecture and engine designed for AI agents, multi-agent workflows, and local LLM research systems.
Apr 2026 · Project
Trisynapse
Desktop AI Research Workspace
A local-first, privacy-focused desktop AI research assistant. Bring your own keys to compile sources into local wikis, explore concept graphs, perform semantic search, utilize an infinite whiteboard canvas, and monitor topics offline.
Feb 2026 · Project
DotChat (SLM Framework)
Causal Language Model & ChatUI
A minimalist PyTorch framework and web interface for training small language models (DotLM) from scratch. Supports pretraining, SFT, DPO, and reasoning (Chain-of-Thought), with a hardware-optimized inference engine using CUDA Graphs and static KV cache.
Feb 2026 · Community
ArxiViz
Neural Network Visualizer
An interactive platform for exploring state-of-the-art neural network architectures across CV, NLP, Generative AI, RL, and GNNs, combining architecture diagrams, mathematical concepts, and PyTorch implementations in a unified learning interface.
Jan 2026 · Dataset
SimpleThoughts Dataset for LLMs
Hugging Face Dataset
A complete synthetic training corpus for LLMs spanning all four training stages, Pretraining, Supervised Fine-tuning (SFT), Preference alignment (DPO), and Reasoning built entirely around simple thought experiments.