System Online • Shubham Gupta

I Build

> AI Systems Engineer
> Agentic AI & Multi-Agent Systems
> LLM Infrastructure, RAG & Vector DBs
> GenAI, LLMOps & Deployment Pipelines
> MLOps, CI/CD & Production Workflows
> Computer Vision & Multimodal Models

MODEL_STATE: ACTIVE
LATENCY: 12ms

// INIT_SYSTEM: Who Am I

Systems Thinker.
Building Intelligent Systems End-to-End.

I specialize in designing and deploying AI systems that operate reliably in real-world environments. My work focuses on combining machine learning, LLMs, and scalable infrastructure to create solutions that are not just accurate, but production-ready. I approach problems from a systems perspective—optimizing across data, models, and deployment to deliver measurable impact.

End-to-End Ownership

From Problem → Model → Deployment → Monitoring

Multi-Domain Experience

Computer Vision • LLMs • Automation Systems

Production-Focused

Deployed, Tested, and Iterated in Real Environments

Optimization Mindset

Improving Accuracy, Latency, and System Efficiency

TECHNICAL_STACK_OVERRIDE

Technical Skills

AI / ML & GenAI

Machine Learning
Deep Learning
Generative AI
Large Language Models (LLMs)
Agentic AI
Autonomous Workflows,Multi-Agent Systems
Retrieval
RAG Systems
Computer Vision
OCR (TrOCR, rOCR)
Optimization
Reinforcement Learning,Genetic Algorithms

Frameworks & Libraries

LLM / Agents
LangChain,LangGraph,AutoGen
Deep Learning
PyTorch,TensorFlow,Keras
ML & Data
Scikit-Learn,Pandas,NumPy
CV & Viz
OpenCV,Matplotlib,Seaborn
NLP
Hugging Face Transformers

LLM & Data Infra

Vector DBs
FAISS,Pinecone,ChromaDB
Knowledge
RAG Pipelines,Retrieval Systems
LLM APIs
OpenAI API
Search
Embedding,Semantic Search

MLOps & Deployment

Tools
Docker,MLflow,Jenkins
APIs
FastAPI,Streamlit
Workflows
Model Deployment,API Dev
Automation
Tracking,Pipeline Auto

Cloud, DevOps & Data

AWS
EC2,S3,Lambda
Ops
Git,CI/CD,Bash
DBs
MySQL,PostgreSQL,MongoDB

Programming

Primary
Python
Secondary
JavaScript,SQL

// SYSTEM_MODULES

Featured Deployments

8
Core APIs
LLM Systems

RAG Playground

Built an interactive Retrieval-Augmented Generation playground with visual node execution, OpenRouter model integration, FAISS retrieval, and embedding-space exploration for rapid RAG experimentation.

FastAPI
React
Vite
FAISS
OpenRouter
99.9%
Uptime
Edge Architecture

Smart Factory Edge MLOps

Designed edge-to-cloud ML infrastructure for industrial IoT. Automated model deployment from cloud to edge devices via NVIDIA Triton and TensorRT.

TensorRT
Kafka
FastAPI
Docker
AWS IoT
2x
Speedup
Generative AI

Document-to-JSON LLM Parser

Built an end-to-end LLM processing pipeline wrapping open-source models (Llama 2) into an API that extracts structured JSON data from raw PDFs.

LangChain
Llama 2
FAISS
Celery
Redis
89%
Accuracy
Computer Vision

Semiconductor Defect Detection

Implemented high-speed real-time defect anomaly detection using a hybrid CNN approach, achieving state-of-the-art precision in minimal ms/inference.

PyTorch
OpenCV
CUDA
YOLOv8
60%
QA Reduction
Vision-Language

Engineering Drawing QA

Automated manual QA checks of complex engineering blueprints via fine-tuned TrOCR + YOLO architectures.

TrOCR
HuggingFace
FastAPI
Vue.js
30fps
Inference
Edge CV

Real-Time Pothole Detection

Deployed lightweight models onto Edge TPUs for real-time video stream processing from mobile vehicles.

TensorFlow Lite
Edge TPU
GStreamer

// EXECUTION_LOG

System Timeline

May 2023 - Jan 2026

AI Systems Engineer

HL Mando

Led AI optimization workflows and automated manual quality checks using Computer Vision. Designed multi-cloud pipelines for deploying optimized models globally. Applied Genetic Algorithms and Reinforcement Learning for component optimization.

CV AutomationGA + RLProduction DeploymentMulti-Cloud
Feb 2022 - Jul 2022

Security Automation Engineer

BreachLock

Architected security automation pipelines. Developed language models for automated vulnerability reporting, reducing manual report generation time significantly.

LLMsPipeline AutomationCybersecurity AI
Aug 2021 - Jan 2021

AI Intern

ThinkingStack

Built YOLO-based Automatic Number Plate Recognition (ANPR) systems and developed highly scalable edge nodes for real-time processing.

YOLOANPREdge AI
shubham@ai-sys:~
Shubham OS v1.0.0 (Linux kernel 6.x)
Type 'help' to see available commands.
>

// RECOGNITION

Awards & Milestones

HL Global Excellence Award

2nd Prize IRTC Research

Spotlight Award

Best Project Award

MSI Contribution Award

Ready to Architect
Custom Intelligence?

Whether you need scalable LLM infrastructure, edge CV models, or a fully connected autonomous pipeline, my systems are ready. Let's connect.

> choose_communication_channel --mode="direct_message"