Engineering Intelligent Solutions

AI & Machine Learning Engineer with 10 months of experience building Computer Vision and Multimodal RAG systems. Engineered end-to-end ML pipelines using Python, PyTorch, and FastAPI.

View Projects Download Resume

Shubham Kumar

AI & Data Engineering Specialist

I am a passionate engineer from Buxar, Bihar with a B.Tech from Lovely Professional University. My professional philosophy is rooted in curiosity and analytical problem-solving.

Engineered end-to-end ML pipelines using Python, PyTorch, and FastAPI, optimizing workflows to reduce manual tasks. Proficient in integrating complex APIs and delivering reliable, scalable backend solutions.

Education

Lovely Professional University

B.Tech - Computer Science and Engineering

Phagwara, Punjab, India | CGPA: 8.50

Sep’2021 – July’2025

Technical Skills

Languages

Python, SQL (MySQL), HTML

Frameworks

PyTorch, FastAPI, Flask, YOLOv8, Docker, AWS (S3)

Tools

Git/GitHub, Postman, Pinecone, Streamlit

Core Competencies

Computer Vision, RAG Pipelines, Object Detection, Prompt Engineering, Multimodal Data Extraction

Experience

Feb 2025 – Nov 2025

Data Engineering & AI/ML Intern

BiUP Technologies Pvt. Ltd., Gurugram

Multimodal RAG Pipeline: Architected ingestion pipeline (PDF, DOCX) using FastAPI & LLMs. Automated feature mapping and Pinecone vector storage, reducing manual data extraction time by 90%.
Real-Time Computer Vision: Engineered vehicle photography guidance using YOLOv8 with <60ms latency. Developed custom IoU metrics and silhouette matching.
MLOps: Enhanced production stability by resolving critical PyTorch serialization issues. Established comprehensive logging and error-handling protocols for integrations.

Featured Projects

Advanced Predictive Maintenance

Full-stack ML platform for aircraft engine health monitoring. Achieved 70.36% R² accuracy on NASA C-MAPSS dataset using XGBoost with Optuna optimization.

XGBoostOptuna FastAPIDocker

Code

Real-Time IoT Intrusion Detection

End-to-end MLOps pipeline detecting DDoS and Botnet attacks with 99.81% accuracy and <50ms latency. Handles extreme class imbalance using stratified sampling.

MLOpsCI/CD XGBoostFastAPI

Code

ProfitGenAI: E-commerce Agent

Hybrid recommendation engine for a 50k+ product catalog. Integrates Llama-based LLM via Groq API to generate persona-driven sales pitches with <150ms response time.

GenAIRAG Groq APIFAISS

Code Demo

Volunteering

Girnar Foundation

Participated in Indoor Plantation Drive to promote urban greenery.

Geeks for Geeks LPU

Participated and Coordinated in different events for the GFG LPU Page.

Get In Touch

Let's Connect

Open for opportunities in Data Science and AI/ML. Feel free to reach out via email or phone.

shubham31103@gmail.com +91 6204992324 LinkedIn Profile GitHub Profile