Loading worlds
Claymation hero world of Hardik Bansal's portfolio: warm clouds with a cat, camera, notebooks and a plant floating in the air
AI and machine learning world: a glowing purple AI model surrounded by floating data cubes, with a cat on a stack of books
Data analytics world: a clay wall of charts and dashboards with stacked data blocks and a cat
Generative AI world: open books turning into miniature 3D worlds with mountains, castles and waterfalls
Retrieval-augmented generation world: a towering library where glowing documents flow into a glass knowledge core
Agentic AI world: a workshop desk surrounded by small clay robot agents, each holding a different tool
Tech stack world: floating clay tiles for Python, databases, charts, cloud and the command line
About and contact world: a rooftop ledge with a lantern, bag and a cat looking over a glowing city at sunset

01 — I'm Hardik Bansal

I build with AI.

AI/ML · GenAI · LLMs · Software Engineering

From computer vision models to LLM-powered workflows, I build AI applications and explore the engineering behind intelligent software.

02 — AI / ML

Teaching machines to spot fakes.

Project · Computer Vision

SyntheticFace Detector

A PyTorch classifier that distinguishes real and manipulated faces using an EfficientNet-B0 model.

  • 58,869 images
  • FaceForensics++ C23
  • EfficientNet-B0 · PyTorch
  • MTCNN · OpenCV face preprocessing
Live Demo View on GitHub

03 — Data & Analytics

From raw data to insight.

Project · Data Analytics · Cloud

InsightFlow

A data analytics and reporting project connecting Python, SQL, cloud storage, and Power BI to turn processed data into useful insights.

  • Python · Pandas
  • MySQL · AWS S3
  • Power BI · Reporting

04 — Generative AI

Making LLMs useful.

Exploring how prompts, large language models, and application logic turn generative AI into useful software.

Currently Exploring

  • LLMs
  • Prompt Engineering
  • LangChain

05 — RAG & Knowledge

Give LLMs the right context.

Exploring embeddings and vector search to retrieve relevant information and ground LLM responses in external knowledge.

Currently Exploring

  • RAG
  • Embeddings
  • Vector Search

06 — Agentic AI

From failed pipelines to answers.

Project in progress

GenCase Automation

An AI-assisted diagnostic workflow designed to analyze failed CI/CD logs, explain potential causes, and email structured reports.

  • Failure log analysis
  • LLM-assisted diagnosis
  • Automated reporting

07 — Tech Stack

Tools behind the work.

Programming
Python · Java · SQL · DSA
Machine Learning & Vision
PyTorch · EfficientNet · OpenCV · MTCNN
Data & Databases
Pandas · NumPy · Power BI · MySQL · PostgreSQL
Cloud & Engineering
AWS S3 · Docker · Git · Linux · CI/CD
Generative AI exploring
LLMs · LangChain · LangGraph · RAG · AI Agents

08 — About / Contact

Let's build something useful.

B.Tech CSE student building at the intersection of AI and software engineering. Interested in AI/ML, GenAI, and intelligent automation. Open to internships, collaborations, and meaningful engineering projects.

GitHub LeetCode