Graduated B.Tech IT • Available for Roles

AI/ML & Systems

Engineering Autonomous Agents & Neural Systems.

I’m Aryan Gupta — an AI/ML engineer focused on Multi-Agent Reinforcement Learning, Computer Vision & Steganography, and scalable predictive architectures. I build systems that coordinate, reason, and operate under high-pressure constraints.

Explore Projects & Lab +91-9474584883 Resume PDF

95%

Seeker Win Rate in 3v3 MARL (100-ep benchmark)

30M+

Timesteps trained via PPO & CUDA Acceleration

-20%

Passenger wait times in MADDPG Fleet Simulation

aryan_gupta@core-system:~$

LIVE_SYSTEM

> "name": "Aryan Gupta"

> "education": "B.Tech Information Technology, Manipal University Jaipur"

> "status": "Graduated — Ready to deploy"

> "research_focus": ["Multi-Agent RL", "Steganography QR", "Fleet Dynamics"]

> "internships": ["MeitY (AI & Security)", "IBM Edunet (AI & Cloud)"]

> "primary_stack": ["Python", "PyTorch", "CUDA", "scikit-learn", "Flask"]

Professional Track Record

Research & Industry Experience

Hands-on contributions in Government Cybersecurity Research and Enterprise Cloud Machine Learning.

AI and Security Research Intern

Ministry of Electronics and Information Technology (MeitY)

June 2025 – July 2025

Utilized Stable Diffusion and ControlNet (computer vision) to generate aesthetic, full-image blended error-correction QR codes, enabling cybersecurity applications like steganography, covert data transmission, and anti-phishing authentication.
Researched and integrated dual-value QR codes with distance-based, cryptographically signed payloads, supporting tiered access control, multi-factor authentication, anti-counterfeiting, and defensive threat modeling.

Interactive MeitY Steganography & Dual-Value Cryptographic Reveal

Drag to Decode Payload

Layer 01: Visual Aesthetic QR (ControlNet + SD)

Slide right to simulate distance-based optical threshold inspection and cryptographically signed covert payload extraction.

Visual Layer (0%) Dual Threshold (50%) ECDSA Signed (100%)

PAYLOAD: blended-artistic-matrix::error_corr_level_H

Stable Diffusion ControlNet Steganography Cryptographic Payloads Defensive Threat Modeling Computer Vision

AI and Cloud Intern

Edunet Foundation, in collaboration with IBM

July 2024 – Aug 2024

Cleaned and preprocessed large datasets — addressing missing values and inconsistencies to ensure data quality and integrity — then developed a robust ARIMA-based time-series forecasting model to drive predictive insights.
Deployed the model on the IBM Cloud platform, leveraging cloud-based infrastructure to enable scalable, efficient model deployment and gaining hands-on experience with cloud technologies.

📈 ARIMA(p,d,q) Time-Series Forecasting & Confidence Interval (IBM Cloud) 95% Confidence Band

ARIMA Forecasting IBM Cloud Data Preprocessing Time-Series Analysis Python Pandas / NumPy

Interactive Autonomous Systems Lab

Engineered Projects & Simulations

Real-time multi-agent simulations, fleet coordination models, predictive decision trees, and media automation pipelines.

Flagship: 3 Seekers (Red) vs 3 Evaders (Cyan)

PPO Policy Active

SEEKERS: 3 EVADERS: 3 TAGS: 0

TIMESTEP: 30,000,000 WIN RATE: 95%

💡 Click anywhere in the arena to drop an interactive disturbance beacon!

Pymunk Physics + CUDA Subprocesses

Reinforcement Learning Jan 2025 – April 2025

Multi-Agent Vehicle Fleet Optimization System

Developed a multi-agent taxi simulation system using Python, PyGame, and PyTorch implementing both A* pathfinding and MADDPG (Multi-Agent Deep Deterministic Policy Gradient) approaches. Reduced passenger wait times by 20% and optimized fleet coordination for 100+ simulated passengers.

3.2 min (-20% Wait Time) 94.8% Fleet Utilization

MADDPG A* Pathfinding PyGame PyTorch Fleet Coordination

Supervised Learning June 2024 – July 2024

Student Performance Prediction

Built a Random Forest Classifier to predict student performance with 96% accuracy. Led end-to-end data preprocessing (imputation, encoding) and hyperparameter optimization using GridSearchCV, boosting model accuracy by 10% and enabling actionable educational interventions.

Study Hours: 18 hrs/wk

Attendance Rate: 88%

Mock Assessment: 82/100

Predicted: High Distinction (A+) 96.0% Model Confidence

Random Forest (96%) GridSearchCV scikit-learn Data Imputation Feature Engineering

Automation & Media Synthesis June 2021 – July 2021

Automated Video Generation Project

Automated end-to-end video creation from Reddit data using Python, API integration, and MoviePy. Reduced video production time by 90% and increased YouTube engagement by 20% through automated workflow optimization and dynamic media composition.

01. Reddit API

Data Ingestion

➔

02. TTS / Script

Dynamic Synthesis

➔

03. MoviePy

Programmatic Render

➔

04. YouTube

+20% Engagement

-90% Production Time +20% Engagement MoviePy Reddit API Python Automation

Technical Competencies

Skills & Domains Universe

Every competency listed here is derived directly from my resume and backed by real project implementations.

Reinforcement Learning Specialty

PPO, MADDPG, Multi-Agent Self-Play

Computer Vision Research

Stable Diffusion, ControlNet, Stego QR

Time-Series Forecasting IBM Cloud

ARIMA Modeling, Predictive Analysis

Supervised Learning 96% Acc

Random Forest, GridSearchCV

CUDA / GPU Acceleration Hardware

GPU Training, 8 Subprocess Envs

Python Proficient

Primary Research & Engineering Lang

C / C++ Systems

Algorithms & High-Performance Compute

JavaScript Familiar

Web Technologies & Visualizations

PyTorch Core ML

Actor-Critic, Policy Gradients, Tensors

scikit-learn Modeling

Pipelines, Ensembles, GridSearchCV

Keras Deep Learning

Neural Architecture Prototyping

NumPy & Pandas Data Engineering

Vectorization, Imputation, Preprocessing

Matplotlib Visualization

Loss Surfaces, Telemetry, Distributions

Pygame & Pymunk Simulation

2D Physics, Rigid Bodies, Raycasts

IBM Cloud Cloud

Model Deployment & Scalable Compute

Weights & Biases MLOps

Hyperparameter Tuning & Logging

Git / GitHub VCS

Branching, PRs, Version Control

Flask Microservices

Model Inference Endpoints & REST

MySQL Database

RDBMS, Relational Schemas, Optimization

Unix / Linux OS / Shell

Shell Scripting, Subprocesses, Servers

Privacy & Security Cybersecurity

Steganography, Threat Modeling, Payloads

Networking Protocols

Data Communication, Covert Transmission

Automation Efficiency

API Integration, Workflow Optimization

Problem-Solving Core

Algorithm Design, High-Pressure Decisions

Academic Foundation

Education & Core Coursework

Formal training in Computer Science, Distributed Systems, and Algorithmic Complexity.

B.Tech in Information Technology

Graduated

Manipal University Jaipur

Aug 2022 – May 2026

RELEVANT RIGOROUS COURSEWORK:

Data Structures & Algorithms Object-Oriented Programming Operating Systems Relational Database Management (RDBMS) Data Communication Computer System Architecture Web Technologies

Science Stream

Completed

The Air Force School, Subroto Park

Mar 2019 – Feb 2022

Foundational background in Physics, Chemistry, and Advanced Mathematics, cultivating deep analytical problem-solving skills.

Impact & Community

Leadership & Extra-Curricular

Leading engineering teams, driving technical research initiatives, and competing in hackathons.

Head of Projects & Research

90% Completion

ACM MUJ Student Chapter

Spearheaded an innovative project team of 15+ students, built by interviewing 40+ applicants, and mentored them to an industry-grade 90% project completion rate across complex software deliverables.

Research Team Member

CyberSpace College Club

Actively contributed to cybersecurity research initiatives, collaborated on technical security projects, and gained practical experience with emerging defensive technologies and security tools.

Hackathon Team Director & Lead

Competitive Hackathons

Directed multidisciplinary hackathon teams, leading end-to-end project development, cross-functional coordination, and successful pitch presentations to expert judging panels under high-pressure competitive environments.