―――――――――――――――――――――――――――――――――――――――――――――|
I build practical AI systems, not just notebooks — retrieval pipelines that actually retrieve, chatbots that actually help, dashboards people actually open twice. I spent a term as VP of my college's AI & Data Science Student Association, and interned at HiveTech getting my first real taste of shipping ML that other people depend on. I'm not here to look busy. Every project below is something I deployed, debugged at 1 AM, or rebuilt because the first version wasn't good enough. role: AI/ML Engineer · GenAI Engineer · Python Developer
based_in: open to relocation, remote, and international roles
status: open to work, immediate joiner
currently: sharpening DSA + shipping one more GenAI project |
me, probably mid-debug |
―――――――――――――――――――――――――――――――――――――――――――――
―――――――――――――――――――――――――――――――――――――――――――――→ DocuQuery — live RAG pipeline, deployed
Ask real questions of real documents. LangChain + FAISS for retrieval, HuggingFace embeddings, Groq's LLaMA3 for generation — running live on Streamlit Cloud after I chased down a broken import path and a deprecated model at midnight. This is the link I lead with in every application.
LangChain FAISS Groq Streamlit
→ ProjectFlow — full-stack SaaS
A project-management SaaS built end to end: Next.js and TypeScript on the front, Prisma and PostgreSQL underneath, deployed on AWS. Built as a virtual internship submission and treated like production code anyway.
Next.js TypeScript Prisma AWS
→ TalentIQ — AI-powered HR platform
A recruitment platform concept exploring where GenAI actually saves recruiters time, versus where it just adds noise.
Python GenAI HR-Tech
→ MindEase — NLP wellness chatbot
A conversational agent designed around empathy, not just intent-matching — built to hold a supportive conversation, not just answer one.
Python NLP Chatbot
→ EnterpriseAI Assistant — self-hosted LLM assistant
DB-GPT running locally in Docker, wired to a SiliconFlow API key — exploring what an enterprise-grade AI assistant looks like outside a vendor's walled garden.
Docker DB-GPT LLMs
→ VAE Playground — generative modeling, from scratch
A Variational Autoencoder built in PyTorch to actually understand the math behind generative models instead of just importing one.
PyTorch Generative AI
――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――Currently learning Spanish, badly. Danced long before I coded. Somewhere between the two I picked up a cat named Moggu, who has strong opinions about my keyboard.
Hiring, or just want to talk shop about RAG pipelines and retrieval hacks?
madhura-23 · open to remote & international roles