STATUS: AVAILABLE FOR 2026 AI / SYSTEMS ROLES
Aryan Gupta

Fluid Intelligence in Motion

AI & Reinforcement Learning Engineer. Designing emergent multi-agent self-play systems, deep computer vision, and cryptographic steganography.

30M+
RL Trajectory Steps
95.4%
Evasion Victory Rate
< 10ms
Inference Latency
SIM_ENV // CUDA:0
ONLINE
ALGORITHM PPO v2.4 (GAE)
PARALLEL ENVS 8x Vectorized
CONVERGENCE 95.4% Rate
STEG SNR +34.2 dB
LIDAR 102-RAY RADIAL TOPOLOGY STEP: 29,481,200
X: 412.89 | Y: -84.21 | θ: 1.48 rad
DISCOUNTED RETURN (γ=0.99) +482.4
memory VRAM: 6.8 / 16.0 GB LOSS: 0.0041 (GAE-λ)
hub Core Competency Vectors

Harmonizing Complexity with Luminous Clarity

Architectural depth spanning probabilistic reinforcement learning agents, perceptual deep neural networks, and fault-tolerant cloud pipelines.

smart_toy

Autonomous Agents (PPO/MADDPG)

Self-play multi-agent dynamics with continuous Pymunk physics, GPU parallelization, and custom reward engineering. Optimized for adversarial and cooperative policy stability.

PPO MADDPG Pymunk CUDA
security

Computer Vision & Security

Aesthetic steganographic QR codes utilizing Stable Diffusion and ControlNet with distance-authenticated cryptographic signatures and generative artifact mitigation.

Stable Diffusion ControlNet Cryptography Latent CV
cloud

Cloud & Predictive AI

Enterprise time-series forecasting (ARIMA) deployed to auto-scaling IBM Cloud environments with automated inference drift monitoring and container orchestration.

ARIMA IBM Cloud REST Microservices Time-Series
science Institutional Research & Directorship

Fellowships & Engineering Leadership

National Frontier R&D

Ministry of Electronics and Information Technology (MeitY)

AI and Security Research Intern

GOVERNMENT OF INDIA

Pioneered generative error-corrected QR steganography using Stable Diffusion and ControlNet for defensive covert communications. Formulated mathematically sound loss bounds balancing perceptual quality and barcode decodability.

Defensive covert transmission protocol
Distance-authenticated cryptographic verification
Enterprise Cloud AI

Edunet Foundation in collaboration with IBM

AI and Cloud Intern

INDUSTRY COHORT

Engineered predictive ARIMA time-series models and deployed production-grade REST endpoints on IBM Cloud infrastructure with containerized pipelines for high-throughput automated forecasting.

Automated model drift pipelines
Production Kubernetes microservice
Leadership & Governance

ACM MUJ Student Chapter

Head of Projects and Research Team

ASSOCIATION FOR COMPUTING MACHINERY

Directed 15+ student developers selected from 40+ applicants, maintaining a 90% project completion benchmark through iterative agile code reviews, architecture critiques, and sprint cadences.

15+ developers mentored
90% delivery rate
4 core research sprints
terminal Deployed Architectures

Selected Implementations

High-performance distributed learning systems, multi-agent coordinate networks, and automated synthesis engines.

SYSTEM 01 // MULTI-AGENT RL sports_esports

MARL Tag: 3v3 Pursuit-Evasion

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

PyTorch PPO + GAE 30M Steps 102-Ray LiDAR 95% Win Rate
SYSTEM 02 // AUTONOMOUS LOGISTICS commute

Multi-Agent Fleet Transit Optimizer

Autonomous taxi simulation combining Multi-Agent Deep Deterministic Policy Gradients (MADDPG) with dynamic A* pathfinding. Slashed passenger wait times by 20% across 100+ simultaneously simulated agents in congested grids.

MADDPG Dynamic A* 100+ Agents -20% Wait Latency
SYSTEM 03 // PREDICTIVE ML analytics

Student Performance Diagnostic

Random Forest classifier achieving 96% evaluation accuracy; optimized via rigorous hyperparameter tuning (GridSearchCV), feature importance factorization, and multi-variable normalization for proactive educational intervention.

Scikit-learn Random Forest 96% Accuracy GridSearchCV
SYSTEM 04 // PIPELINE AUTOMATION video_settings

Automated Video Generation Engine

End-to-end Python engine converting structured metadata and real-time feeds into programmatic video assets using MoviePy and REST endpoints, reducing manual rendering latency by 90%.

Python MoviePy REST API -90% Latency
grid_view Technical Stack Diagnostics

Computational & Engineering Matrix

code Languages & Core Systems

Python (PyTorch / RL Stack) EXPERT // 95%
CUDA Acceleration & Parallel Compute ADVANCED // 88%
C / C++ (Algorithmic Core) PROFICIENT // 82%
Unix / Linux & Kernel Optimization EXPERT // 90%
SQL & Relational Schemas SOLID // 85%

psychology AI, Frameworks & Environments

PyTorch & Deep Learning EXPERT // 96%
Pymunk & 2D Physics Simulators ADVANCED // 90%
scikit-learn & Statistical Inference EXPERT // 92%
Stable Diffusion & ControlNet Latents ADVANCED // 88%
NumPy, Pandas & Matplotlib EXPERT // 95%
ADDITIONAL ARSENAL: Docker Git & CI/CD IBM Cloud API RESTful Services PyGame Keras TensorBoard
radar Channel Transmission

Connect with Aryan Gupta

Available for 2026 roles in AI, Machine Learning, and Systems Engineering. Whether discussing emergent multi-agent reinforcement learning topologies or production deployment pipelines, reach out directly.