CURATED STYLES:
自然 — ORGANIC INTELLIGENCE

Simplicity, Imperfection, & Emergent Order.

I am an AI and reinforcement learning engineer cultivating multi-agent self-play systems, computer vision, and defensive architectures. Grounded in patience, computational harmony, and unadorned precision.

30M+ Timesteps

Continuous self-play convergence without policy degeneration.

95% Benchmark

Seeker equilibrium achieved in pursuit-evasion benchmarks.

01 // PHILOSOPHY

The Beauty of Unforced Equilibrium.

Autonomous Agents

Training multi-agent actors with PPO and MADDPG, allowing optimal coordination to emerge organically through environment friction.

Steganographic Vision

Utilizing Stable Diffusion and ControlNet at MeitY to embed subtle dual-value cryptographic signatures invisible to the restless observer.

Cloud & Forecasting

Deploying ARIMA time-series models onto scalable IBM Cloud infrastructure with clean, calm architectural minimalism.

02 // FELLOWSHIPS & JOURNEY

A Chronicle of Dedicated Craft.

Ministry of Electronics and Information Technology (MeitY)

AI and Security Research Intern

Researched aesthetic error-corrected QR steganography utilizing Stable Diffusion and ControlNet. Designed distance-based cryptographically signed payloads supporting tiered access authentication.

Edunet Foundation in collaboration with IBM

AI and Cloud Intern

Trained predictive ARIMA time-series forecasting pipelines and served containerized endpoints on IBM Cloud infrastructure.

ACM MUJ Student Chapter

Head of Projects and Research Team

Mentored an engineering group of 15+ student developers to a 90% project delivery benchmark, fostering mindful collaboration.

03 // SELECTED WORKS

Creations in Code.

I

MARL Tag: 3v3 Pursuit-Evasion

PyTorch self-play reinforcement learning system utilizing PPO with GAE. Parallelized over 8 CUDA environments (30M steps) with 102-ray perception and 95% victory rate.

PyTorchPPOCUDAPymunk
II

Multi-Agent Fleet Transit Optimizer

Autonomous taxi simulation combining MADDPG with dynamic A* pathfinding. Slashed passenger wait times by 20% across 100+ concurrent simulated entities.

MADDPGA*PyGamePyTorch
III

Student Performance Diagnostic

Random Forest classifier with 96% accuracy; optimized via GridSearchCV and rigorous data normalization for educational intervention.

Random Forestscikit-learnPandas
IV

Automated Video Generation Engine

End-to-end Python pipeline converting structured data into automated video media using MoviePy and REST APIs, reducing manual latency by 90%.

PythonMoviePyAutomation
04 // DISCIPLINES

Technical Arsenal.

Languages & Core Systems

Python (Proficient), C/C++, JavaScript, SQL, CUDA GPU Acceleration, Linux.

AI & Reinforcement Learning

PyTorch, scikit-learn, Keras, NumPy, Pandas, PPO, MADDPG, GAE, Self-Play Dynamics, Pymunk.

Vision & Cyber Security

Stable Diffusion, ControlNet, Cryptographic Verification, Error-Correction QR, OpenCV.

Academic Foundation

B.Tech in Information Technology, Manipal University Jaipur (Aug 2022 – May 2026).

05 // DIALOGUE

Open to Mindful Collaborations.

Available for engineering roles in 2026. Reach out directly via email or telephone.