DΞVΛΠIK's AI Research Ecosystem

Scalable Neuro-Symbolic Architectures & Autonomous Agent Systems

---

"Investigating the mechanisms of internal self-model formation and metacognition in neural architectures."

---

📖 Table of Contents

• About the Researcher
• Research Philosophy
• Featured Breakthrough Projects
• Research Domains
  • Metacognition & Cognitive Architecture
  • Reinforcement Learning & Game Theory
  • Generative AI & Diffusion Systems
  • Large Language Models & Agents
  • Computer Vision & Image Processing
  • Astrophysics & Computational Cosmology
  • Retrieval-Augmented Generation & Memory Systems
  • Neural Architecture & Theory
  • Production Applications & Tools
• Learning Resources & Community Contributions
• Achievements & Recognition
• Research Trajectory
• Collaboration & Contact

---

🎓 About the Researcher

Devanik
B.Tech in Electronics & Communication Engineering '26
National Institute of Technology Agartala

Fellowships & Affiliations:

• 🏅 Samsung Convergence Software Fellowship (Grade I) - Indian Institute of Science
• 🚀 ISRO Space Hackathon Winner - National-level space technology innovation
• 🔬 Research Intern (Astrophysics × ML) - Interdisciplinary computational cosmology

Core Technical Competencies:

• Cognitive Engineering - Designing high-reliability architectures for internal state estimation
• Applied Mathematics - Information Geometry, Causal Inference, and Topological Data Analysis (TDA)
• Distributed Systems - Scalable memory architectures for massive-agent simulations
• Bio-Inspired Computing - Neuromorphic plasticity rules and homeostatic regulation algorithms

---

🧬 Research Philosophy

Systems-Theoretic Approach

My work focuses on designing high-reliability cognitive architectures and mathematically grounded self-organizing systems.

This portfolio represents a rigorous engineering effort in distributed state estimation, control-theoretic stability, and scalable neuro-symbolic integration. Each project targets specific bottlenecks in current AI systems—memory consolidation, causal reasoning, and energetic efficiency.

Engineering Methodology:

1. Metric-Driven Development - Quantifying internal state dynamics via information geometry.
2. Scalability First - Designing architectures that survive orders-of-magnitude scaling.
3. Reproducible Emergence - Rigorous benchmarking of emergent behaviors under adversarial conditions.
4. Mathematical Grounding - Leveraging information theory (entropy, mutual information) as the primary optimization objective.

---

🌟 Featured Breakthrough Projects

🔮 Divine Monad Experimental Exploration of Causal Emergence ✨ Algorithm: Recursive Causal Inference Engine 🧠 Key Metric: Effective Information (EI) & Integrated Information (Φ) 📊 Architecture: Autopoietic feedback loops for self-regulation 🔬 Validation: Intervention-based protocols for quantifying causal structure Technical Contribution: Implements quantifiable metrics for higher-order information integration in distributed systems.	🌊 Recursive Hebbian Organism Neuromorphic Continual Learning System 🔄 Architecture: 21-stage Neuromorphic Continual Learning Framework 🌱 Mechanism: Hebbian Plasticity with Homeostatic Regulation 🧬 Optimization: Mitigating Catastrophic Forgetting via replay buffers 🎯 Objective: Meta-learning stable representations across non-stationary tasks Technical Contribution: Demonstrates high-stability learning in curriculum-free environments.
🧬 AION: Algorithmic Reversal of Genomic Entropy Longevity Research via Information Theory ⏳ Treats aging as entropy accumulation in genetic networks 🔬 AI-driven interventions for epigenetic reprogramming 📈 Predictive modeling of cellular senescence pathways 🧪 Integration of systems biology with deep learning Impact: Novel computational approach to lifespan extension	🌌 Lucid Dark Dreamer Neural Dream Consolidation Mechanisms 💤 Models REM/non-REM memory consolidation cycles 🔮 Counterfactual experience generation without environment interaction 🧠 World model learning through simulated experience 🎨 Latent space imagination for abstract concept formation Impact: Investigates self-supervised representations via offline consolidation.
⚡ Thermodynamic Mind Free Energy Principle Implementation 🌡️ Framework: Active Inference Control System 📉 Objective: Variational Free Energy Minimization (VFE) 🔄 Architecture: Hierarchical Predictive Coding (PC) 🧮 Implementation: Bayesian belief updating in silicon substrates Technical Contribution: Applies the Free Energy Principle to adaptive agent control systems.	🎮 ISRO Mining Site Localization Space Hackathon Winning Project 🛰️ Multi-modal satellite imagery analysis 🗺️ Mineral deposit classification via CNNs 📍 Terrain-aware site optimization 🚀 Production-ready deployment for lunar missions Impact: National recognition for space resource utilization

---

📚 Research Domains

1. 🧠 Technical Competencies & Implementations

Investigating metacognition, state-estimation, and self-modeling architectures

Project	Description	Core Architecture / Algorithm
causa-sui	Researching causal emergence in feedback-heavy architectures	Integrated Information (Φ) as a metric
Recursive Hebbian Organism	Developmental learning with 21 cognitive stages	Neuromorphic continual learning
Thermodynamic Mind	Free energy principle for AI	Active inference framework
Dark Thermodynamic Mind	Extended mind hypothesis in computation	Embodied cognition for robots
Lucid Dark Dreamer	Dream-based memory consolidation	World model imagination
NanoCortex	Miniature cortical column simulation	Efficient neuromorphic processing
Engram	Persistent memory trace formation	Synaptic tagging mechanisms
MateriaMind	Substrate-dependent information integration	Integrated information in materials
MindScopeX	Neural state monitoring and internal diagnostics	Quantifying representation stability
Evolving-AI	Genetic algorithms for architecture search	Self-modifying neural structures
Self-Correcting Language Model	Metacognitive error detection	Confidence-aware generation

2. 🎮 Reinforcement Learning & Game Theory

Mastering strategic reasoning through multi-agent environments and complex games

Project	Description	Core Architecture / Algorithm
AI Chess Nemesis	Deep RL chess engine with AlphaZero-style MCTS	Self-play improvement
MiniChess Deep RL	Compact chess variant for RL research	Reduced complexity analysis
MiniChess RL	Alternative implementation with policy gradients	PPO vs A3C comparison
Checkers RL	Q-learning and TD(λ) for checkers	Temporal difference methods
Sophisticated Sudoku RL	Constraint satisfaction via RL	Graph neural networks
Tic-Tac-Toe RL Battle Arena	Multi-agent tournament platform	Population-based training
RL Super Tic-Tac-Toe	Meta-game with recursive structure	Hierarchical RL policies
The Game of Nim RL	Combinatorial game theory meets RL	Optimal strategy convergence
Qwirkle RL	Pattern recognition board game	Visual feature learning
RL CartPole Environment	Classic control theory benchmark	Policy gradient comparison
Advanced RL Maze Solver	Multi-objective navigation	Curiosity-driven exploration
RubIKSolVeR	Rubik's Cube via deep RL	Admissible heuristic learning
open_spiel	DeepMind's game theory framework (fork)	Multi-agent research toolkit
Game Theory	Nash equilibrium and mechanism design	Economic modeling

3. 🎨 Generative AI & Diffusion Systems

Creating novel content through probabilistic modeling and latent space manipulation

Project	Description	Core Architecture / Algorithm
Fastest Text-to-Image Generator	Optimized Stable Diffusion with <2s inference	Distilled DDPM sampling
text-to-image	CLIP-guided diffusion implementation	Cross-modal conditioning
text-to-image-imagen	Google Imagen architecture recreation	Cascaded super-resolution
image-gen-vercel	Serverless diffusion with Vercel edge	Latent consistency models
vercel-fal-image-generator	Fast API latent generation	Real-time inference
openPipe-ART	Pipeline for artistic style transfer	Multi-stage refinement
Image Playground	Interactive generative art studio	UI for latent exploration
DreamSketch	Sketch-to-image generation	ControlNet conditioning
Text-to-Video Generator	Temporal coherent video synthesis	3D U-Net diffusion
CycleGAN	Unpaired image-to-image translation	Cycle consistency loss
Landscape GAN	Photorealistic terrain generation	Progressive growing GAN
iGAN	Interactive generative adversarial networks	Real-time latent manipulation
GANs	Comprehensive GAN implementations	DCGAN, WGAN, StyleGAN
pytorch-generative-model-collections	VAE, GAN, Flow, Diffusion zoo	Unified training pipeline
MusicVAE	Variational autoencoder for music	Hierarchical LSTM decoder
ht-diffusers	Hugging Face diffusers fork	Custom schedulers
dinov3	Self-supervised vision transformers	DINO v3 implementation
Awesome Text-to-Image	Curated diffusion resources	Research paper collection

4. 🤖 Large Language Models & Agents

Building autonomous systems with reasoning, tool use, and multi-step planning

🧠 Foundation Models & Fine-tuning

Project	Description	Core Architecture / Algorithm
DeepSeek R1	Reinforcement learning from human feedback	Chain-of-thought reasoning
DeepSeek Math V2	Mathematical reasoning specialist	Formal proof generation
Qwen3	Multilingual foundation model (fork)	Cross-lingual transfer
Qwen3 Coder	Code generation and completion	Repository-level context
Qwen Agent	Tool-augmented reasoning framework	Function calling API
Qwen Image	Vision-language model	Interleaved multimodal input
Kimi K2	Long-context LLM (200k tokens)	Efficient attention mechanisms
Gemini	Google Gemini API integration	Native multimodality
nanoGPT	Minimalist GPT training from scratch	Educational implementation
transformers	Hugging Face transformers (fork)	Model hub ecosystem
LLMs from Scratch	Step-by-step LLM construction	Pedagogical resource

🛠️ Agentic Systems & Frameworks

Project	Description	Core Architecture / Algorithm
Agentic RAG R1	Reasoning over retrieved documents	Multi-hop question answering
agent-zero	Minimal autonomous agent framework	Planning and execution
ai-agents-for-beginners	Educational agent development	Step-by-step tutorials
agent-starter-pack-google-gen-AI	5-day intensive course	Production-ready templates
Basic Agent with Tool	Tool-calling LLM wrapper	Function execution
Mini Research Assistant	Search-augmented QA agent	Web scraping integration
AI Search & QA Agents	Information retrieval specialist	Multi-source synthesis

🔗 LangChain & LangGraph Ecosystem

Project	Description	Core Architecture / Algorithm
langchain	LangChain framework (fork)	Chain-of-thought orchestration
LangChain AI Chat App	Conversational agent with memory	Context-aware responses
LangGraph AI Chat App	State machine for complex flows	Graph-based reasoning
langflow	Visual agent builder	No-code LLM pipelines
langgraph-studio	Interactive graph debugger	Workflow visualization
Custom Chatbot using Langchain	Domain-specific assistant	Fine-tuned persona
LLM-powered Chatbot (DSAI Club)	Club management bot	Event coordination

🧰 Gemini & Multimodal Applications

Project	Description	Core Architecture / Algorithm
gemini-cli	Command-line Gemini interface	Terminal integration
run-gemini-cli	Shell script wrapper	Automation ready
Gemini Autocomplete VSCode	IDE code completion	Context-aware suggestions
gemini-genai-processors	Data processing pipelines	Batch inference
gemini_icpc2025	Competitive programming assistant	Algorithm generation
python-genai	Google GenAI Python SDK	API client library
Multimodal Assistant with Langflow	Vision + text reasoning	Cross-modal grounding

🌐 API Integrations & Cookbooks

Project	Description	Core Architecture / Algorithm
openai-cookbook	OpenAI API examples	GPT-4 patterns
anthropic-cookbook	Claude API recipes	Constitutional AI
Chatgpt-4-bot	Discord/Telegram integration	Social platform bots
Perplexity	Search-grounded generation	Citation tracking
Perplexity v2	Enhanced retrieval pipeline	Relevance ranking

📝 Specialized Applications

Project	Description	Use Case
AI Writing Studio	Long-form content generation	Blog/article creation
LinkedIn Post Generator	Professional content writer	Social media marketing
AI Resume and CV Creator	ATS-optimized resume builder	Job application
AI Nano Course Generator	Curriculum design assistant	Educational content
AI Note Generator	Study materials from lectures	Student productivity
PromptWizard	Prompt engineering optimizer	Meta-prompting
AI Daily Creative Prompt Hub	Inspiration generator	Creative writing
AI Dream Interpreter	Symbolic dream analysis	Psychology application
AI Mindmap Generator	Concept relationship mapper	Knowledge structuring
Character AI	Persona-based chatbot	Role-playing dialogue

🔧 Developer Tools & Extensions

Project	Description	Platform
vscode-copilot-chat	GitHub Copilot integration	VS Code
nexus-ai-chat-vscode	Custom AI chat panel	VS Code
nexus-ai-code-assist-vscode	Code completion plugin	VS Code
PixelForge AI Image Generator VSCode	In-editor image generation	VS Code
my-first-extension	VSCode extension template	Educational project

5. 👁️ Computer Vision & Image Processing

Extracting structure and meaning from visual data through deep learning

🖼️ Advanced Vision Systems

Project	Description	Core Architecture / Algorithm
ImageReasoning	Visual question answering	Attention-based reasoning
Yolo Object Detection Web App	Real-time detection API	YOLOv8 deployment
Advanced Star Classification App	Stellar object categorization	ResNet embeddings
Alien Signal Classification	SETI radio signal analysis	CNN spectrogram processing
alphafold-deepmind	Protein structure prediction (fork)	Equivariant transformers
AlphaFold	Alternative implementation	Evolutionary scale modeling
alphafold3	Latest AlphaFold iteration	Diffusion for structures

🎨 Image Processing & Enhancement

Project	Description	Core Architecture / Algorithm
Digital Image Processing Sparse	Compressed sensing reconstruction	L1 minimization
Denoising Images using DSP	Signal processing filters	Wiener/Kalman filtering
Denoising Audio with HPF	High-pass filter design	Butterworth/Chebyshev
The Bible of Digital Image Processing	Comprehensive tutorial collection	Classical algorithms
BG Remover	Background segmentation	U-Net matting
Sophisticated Palette	Color scheme extraction	K-means clustering

📱 Android Computer Vision Apps

Project	Description	Framework
gallery	Photo management with ML	TensorFlow Lite
Simple Gallery	Lightweight image viewer	Kotlin native

6. 🌌 Astrophysics & Computational Cosmology

Applying machine learning to astronomical data and gravitational simulations

Project	Description	Core Architecture / Algorithm
ISRO Mining Site Final App	Space Hackathon Winner - Lunar resource mapping	Satellite imagery analysis
ISRO Space Mining App	Mineral deposit prediction	Multi-spectral classification
Surya IBM × NASA	Solar activity forecasting	Time-series transformers
Cosmic Craftathon	Space exploration simulator	Physics-based engine
Deep Universe	Cosmological parameter estimation	Bayesian neural networks
Black Hole	General relativity simulations	Ray tracing geodesics
Bi-LSTM Light Curve Reconstruction	Stellar variability modeling	Recurrent architectures
AI Cyclone Prediction	Hurricane trajectory forecasting	Spatio-temporal CNNs
Galaxy Dataset	Morphology classification	Transfer learning
evEntHorizoN_Z	Event horizon visualization	Relativistic rendering
QuasarQ	Quasar spectrum analysis	Dimensionality reduction
Life Beyond	Exoplanet habitability scoring	Random forest classifier
SpectraNova	Spectral line identification	Attention mechanisms

7. 📚 Retrieval-Augmented Generation & Memory Systems

Building knowledge-grounded AI with external memory and information retrieval

Project	Description	Core Architecture / Algorithm
Agentic RAG R1	Reasoning over retrieved context	Multi-hop retrieval
rag-flow	Production RAG pipeline	Vector database integration
code-graph-rag	Code repository QA	Graph neural retrieval
Chat With Your Files	Document conversational interface	Semantic chunking
AI Video Insights	YouTube transcript analysis	Temporal grounding
AI Audio Overview	Podcast summarization	Whisper + GPT-4
VisionaryBook	Visual knowledge base	Image-text alignment

8. 🏗️ Neural Architecture & Theory

Exploring novel architectures, training dynamics, and theoretical foundations

Project	Description	Core Architecture / Algorithm
AI Architecture Playground	Experimental model zoo	Custom layer implementations
BSHDER Architecture	Biologically-inspired hierarchy	Sparse coding modules
ANN Zero to Hero	Pedagogical neural network series	Backprop from scratch
DeepSpeed	Microsoft training optimization (fork)	ZeRO memory reduction
tensorflow	TensorFlow framework (fork)	Custom ops development
safe-tensors	Secure model serialization	Rust-based library
vjepa2	Video joint embedding predictive architecture	Self-supervised video
dinov3	Self-distillation with no labels	Vision transformers
Bohrium	High-performance computing backend	GPU acceleration
A Machine Learning Approach for Optimal Power Gating	Hardware-software co-design	Energy efficiency

9. 🚀 Production Applications & Tools

Deployable systems solving real-world problems across diverse domains

💼 Business & Analytics

Project	Description	Tech Stack
E-commerce Product Insights Suite	Customer analytics dashboard	Streamlit + Plotly
AI Stock Insights	Financial market analysis	Yahoo Finance API
Netflix Stocks App	Time-series forecasting	LSTM predictions
Netflix Insights App	Content recommendation engine	Collaborative filtering
Dashboard Creator	Automated visualization generator	Matplotlib/Seaborn
GDP Dashboard App	Macroeconomic indicators	Pandas + Streamlit
Data Analysis App	Automated EDA pipeline	Statistical tests
Stackoverflow Analysis	Developer trends mining	NLP on Q&A data

🌾 Agriculture & Healthcare

Project	Description	Impact
ADV Crop Prediction App	Yield forecasting system	Random forest regressor
ML Deployment Crops	Production agriculture API	FastAPI + Docker
Adv Disease Prediction App	Medical diagnosis assistant	CNN on X-ray/MRI
planto AI Final Project	Plant disease identification	MobileNet transfer learning

🎮 Entertainment & Creative Tools

Project	Description	Category
ChefSynth	Recipe generation from ingredients	NLP + knowledge graphs
AI Weather App	Hyperlocal forecasting	Ensemble weather models
AI NewsVault	Personalized news aggregation	Content-based filtering
Movies Dataset App	Film recommendation system	Matrix factorization
mangle AI Language	Conlang generator	Markov chain grammar
cipher-of-kings	Historical cryptography simulator	Classical ciphers
CryptX	Modern encryption toolkit	AES/RSA implementation

🏢 Workplace & Productivity

Project	Description	Purpose
HRM	Human resource management system	Employee database
CoNnEcTX	Professional networking platform	Graph-based matching
DriftNotes	Markdown note-taking app	Version control integration
QuantumDocs	Collaborative documentation	Real-time editing
Fast API Prometheus Grafana	Monitoring stack setup	Observability platform
Interview Prep Master	Technical interview simulator	LeetCode-style problems
DSA Mastery with Gemma 3	Data structures tutor	Interactive learning
AlgoDiary	Algorithm complexity analyzer	Big-O visualization

📱 Android Applications

Project	Description	Framework
Advanced Robot Monitoring	IoT sensor dashboard	Android + Firebase
AmazeFileManager	Material Design file browser (fork)	Kotlin
MaterialFiles	Alternative file manager	Material You
Olauncher	Minimalist launcher	Simplicity-focused
olauncher-light	Ultra-lightweight launcher	Battery-optimized
TextLauncher	Text-based app launcher	Accessibility

🎨 Creative & Experimental

Project	Description	Core Architecture / Algorithm
Singularity AI	AGI conceptual framework	Thought experiment
void	Minimal text editor philosophy	Digital minimalism
Eerie.EXE	Creepypasta-inspired game	Narrative design
openclaw	Game engine reconstruction	Reverse engineering
Xenofit AI	Alien biology simulation	Procedural generation
open-mind-OM1	Philosophy of mind simulator	Phenomenology exploration

🧪 Research Infrastructure

Project	Description	Use Case
code-samples-for-poc-streamlit	Proof-of-concept templates	Rapid prototyping
EDA Improved	Enhanced exploratory data analysis	Statistical profiling
audio-file-error-handling-gpt-4	Robust audio processing	Exception handling
Postman Challenge	API testing scenarios	QA automation

🎓 Internship & Client Projects

Project	Organization	Role
AI-ML Developer Jayadhi Limited	Jayadhi Limited	ML Engineer
Bharat Intern	Bharat Intern	Data Science
Planto AI Intern	Planto AI	Computer Vision
Coding JR AI Research Docs	Coding JR	Research Documentation
Primetrade.ai Data Science Assignment	Primetrade.ai	Quant Analysis

🏗️ Web Development & Deployment

Project	Description	Stack
e-web	E-commerce platform	MERN stack
Project-X	Full-stack application	React + Django
Foundry-Local	Smart contract testing	Solidity + Hardhat

---

📖 Learning Resources & Community Contributions

Curated educational materials, forks, and collaborative projects

🎓 Foundational Courses

Resource	Provider	Coverage
Microsoft GenAI For Beginners	Microsoft	12-lesson curriculum
AI Foundations DeepMind	DeepMind	Mathematics of AI
ML For Beginners	Microsoft	Classical ML algorithms
Data Science For Beginners	Microsoft	End-to-end pipeline
Python Interactive Textbook	Community	Jupyter notebook tutorials

📚 Advanced Guides & References

Resource	Topic	Depth
awesome-generative-ai-guide	Diffusion & LLMs	Comprehensive bibliography
awesome-datascience	Data Science	Curated tool list
The AI Bible	All AI subfields	Encyclopedic reference
AI Journals	Research papers	Annotated collection
AI Tools	Software ecosystem	Production-ready tools
synthid-text	Watermarking	DeepMind's SynthID

🛠️ Development Practice

Project	Skill Area	Format
Opensource Practice	Git workflows	Contribution guide
Open-source Practice	Collaboration	Pull request training
PythonRobotics	Robotics	Algorithm implementations
My ML Projects	Portfolio	Personal collection

👤 Profile Repositories

Repo	Purpose
Devanik21	GitHub profile README
Debanik21	Alternative profile
Certificates	Achievement documentation
IMP	Important references
markdown	Documentation templates

🤝 Forked Research Tools

Fork	Original Author	Purpose
langchain	LangChain	Custom extensions
transformers	Hugging Face	Model experiments
tensorflow	Google	Custom ops
DeepSpeed	Microsoft	Training optimization
open_spiel	DeepMind	Game theory research

---

🏆 Achievements & Recognition

🎖️ Awards & Fellowships 🥇 ISRO Space Hackathon Winner (2024) National-level space technology competition Lunar mining site localization system Recognized by Indian Space Research Organisation 🎓 Samsung Convergence Software Fellowship (Grade I) Awarded by Indian Institute of Science Top-tier research fellowship for exceptional potential Focus: Bio-inspired cognitive architectures 🔬 Research Intern - Astrophysics × ML Interdisciplinary computational cosmology Applied deep learning to gravitational simulations Publication-track research collaboration

📊 Engineering Impact 📈 257 Active Repositories Comprehensive exploration across 9 major AI domains 50+ production-ready applications deployed 20+ experimental architecture prototypes and research frameworks 🌟 Core Engineering Contributions Recursive Causal Inference Engine (Divine Monad) Neuromorphic Continual Learning Framework (Recursive Hebbian Organism) Genomic Entropy Reversal Algorithms (AION) Neuro-Symbolic Memory Consolidation (Lucid Dark Dreamer) 🎯 Technical Breadth 15+ RL game-playing agents 10+ generative AI systems (diffusion, GAN, VAE) 25+ LLM applications and agentic frameworks 12+ astrophysics ML projects

🧠 Advanced Cognitive Architectures Implemented

1. Divine Monad - Causal emergence investigation with integrated information metrics
2. Recursive Hebbian Organism - Neuromorphic continual learning (21 developmental stages)
3. Thermodynamic Mind - Free energy principle implementation
4. Lucid Dark Dreamer - Dream-based memory consolidation
5. Dark Thermodynamic Mind - Extended mind hypothesis for embodied cognition
6. NanoCortex - Efficient cortical column simulation
7. MateriaMind - Structural information integration in substrates
8. Self-Correcting Language Model - Metacognitive error detection

---

🗺️ Technical Roadmap

🔬 Engineering Priorities (2025-2026)

🎯 Distributed State Estimation Objective: Quantify the informational structure of foundation models via causal emergence metrics. Approach: Measure integrated information (Φ) in Transformers Identify macro-scale abstractions in latent space Test intervention-based structural criteria Extend to multimodal diffusion models Expected Outcome: Quantifiable insights into the macro-scale informational structure of LLMs.

📐 Mathematical Foundations Objective: Formalize control-theoretic bounds for adaptive self-modeling systems. Approach: Information-theoretic self-modeling theorems Causal structure learning algorithms Topological data analysis for internal states Quantum-inspired memory architectures Expected Outcome: Mathematical guarantees on self-model integration and stability.

⚖️ Ethics & Safety Objective: Engineering safety constraints for complex self-modeling agents. Approach: Evaluating markers for moral patienthood in advanced AI Quantifying internal negative utility states in feedback systems Developing ethical guidelines for high-agency systems Safe deployment protocols Expected Outcome: Technical safety recommendations for AGI labs and governments.

🗺️ Long-term Engineering Goals (2026-2030)

Phase 1: System Validation & Benchmarking (2025-2026)
├─ Refining causal emergence metrics in distributed neural systems
├─ Peer-reviewed publications in cognitive engineering
└─ Safety benchmarking of autopoietic architectures

Phase 2: Distributed State Estimation (2026-2027)
├─ Investigating structural markers of information integration in large-scale transformers
├─ Homeostatic control systems for self-regulation
└─ Cross-architecture metabolic and information efficiency metrics

Phase 3: Autonomous Control Systems (2027-2029)
├─ General agency with robust internal model estimation
├─ Control-theoretic guarantees for high-agency systems
└─ Alignment through mathematically grounded self-models

Phase 4: Post-Biological Architectures (2029+)
├─ Information-theoretic substrates for AGI alignment
├─ Human-AI collaborative intelligence
└─ Exploration of non-biological cognitive substrates

🎓 Academic Pathway

Planned Graduate Research (2026+):

• PhD Focus: Computational Theories of Consciousness
• Thesis Direction: Bridging integrated information theory with deep learning
• Publication Goals: NeurIPS, ICML, Nature Neuroscience, JCS (Journal of Consciousness Studies)

---

🤝 Collaboration & Contact

💡 Open to Collaboration On:

• 🧠 Advanced internal modeling - Investigating self-referential loops and metacognition
• 🔬 Neuroscience × AI - Bio-inspired architectures and learning rules
• 🌌 Astrophysics ML - Cosmological simulations and data analysis
• 🎮 Multi-agent RL - Game theory and emergent cooperation
• 🧬 Computational biology - Protein folding, longevity research
• 🤖 AGI safety - Alignment through self-modeling and feedback control

📬 Get in Touch

@Devanik21

Professional Network

@devanik2005

Research Publications

🎯 For AGI Labs & Research Institutions

I'm particularly interested in:

• Research scientist positions focusing on cognitive architectures and self-modeling
• Collaborative projects bridging neuroscience and AI safety
• Opportunities to work on foundation model interpretability and emergent properties
• Interdisciplinary research at the intersection of physics, neuroscience, and ML

What I Bring:

• 257 repositories demonstrating breadth and depth across AI
• Proven track record of experimental architecture prototypes (Divine Monad, Recursive Hebbian Organism, AION)
• National-level recognition (ISRO Hackathon Winner, Samsung Fellowship)
• Strong mathematical foundations and implementation skills
• Passion for investigating the fundamental limits and metrics of cognitive complexity.

---

🌟 Philosophy

"The path to AGI is not through bigger models, but deeper understanding—
systems capable of internal self-model estimation and adaptive confidence tracking."

---

🔬 Computational Approaches to Self-Modeling and Emergence

This repository ecosystem represents a systematic engineering effort. From reinforcement learning agents to thermodynamic minds, each project investigates how higher-order informational patterns and complex self-models can be reliably engineered in distributed systems.

The work continues.

---

Last Updated: February 2026
Status: Active Research
Next Milestone: Causal emergence metrics in large-scale foundation models

---

Built with rigorous engineering, driven by control theory, sustained by the pursuit of scalable intelligence.

© 2026 Devanik | National Institute of Technology Agartala