Hi, I'm Karthik Daivadnya
I build end-to-end ML systems — from predictive maintenance models to RAG pipelines and full-stack dashboards...
Grounded in fundamentals, shipped end-to-end.
I like taking a model past the notebook — into an API, a UI, and a repo someone else could actually run.
I'm a 2026 Computer Science & Engineering graduate from Bapuji Institute of Engineering and Technology, Bangalore, Karnataka — CGPA 8.76.
My focus is building complete systems: a predictive maintenance platform trained on NASA's CMAPSS dataset, a retrieval-augmented document assistant, and ETL pipelines that end in a dashboard someone can actually read. I'm currently interviewing for AI/ML, data science, and software engineering roles.
B.E, Computer Science & Engineering
Flagship project build cycle
Built three end-to-end systems from scratch — predictive maintenance ML, a RAG document assistant, and an ETL + BI pipeline — to anchor internship and new-grad applications.
Job search & take-home assignments
Completed take-home assignments including a legal-contract clause extraction pipeline (CUAD dataset) and a full-stack mini content engine, alongside active interview prep.
Tools I reach for, organized by what they're for.
Note: proficiency bars are intentionally skipped — a specific % here would be a made-up number, not a measured one. Categories reflect real, applied experience.
Programming
AI / ML
Data Science
Web Development
Databases
Tools
Five systems, built and pushed end-to-end.
Every project ships with a real repo. None are deployed to a live URL yet, so demo links open a contact prompt instead of a broken link.
AeroSense
ML platform trained on NASA's CMAPSS turbofan degradation dataset, predicting remaining useful life. Originally built targeting GE Aerospace's Data Science Intern track.
- Random Forest & Gradient Boosting models, R² = 0.975
- FastAPI backend serving live predictions
- React dashboard for fleet-level health monitoring
DocMind
Retrieval-augmented document Q&A assistant built from scratch on LangChain and Mistral AI.
- FAISS vector store for semantic search over uploaded docs
- Upload-and-ask workflow via a FastAPI + React interface
- Resolved a Python 3.14 → 3.11 venv compatibility issue with LangChain
SalesPulse
End-to-end ETL pipeline over realistic raw sales data, ending in dashboards people can actually read.
- SQL transformations into SQLite from raw source data
- Automated Excel report generator
- Interactive Chart.js dashboard + companion Power BI report
Multi-Agent AI System
Five cooperating agents orchestrated with LangGraph, equipped with tools and persistent memory.
- 5 agents coordinated via LangGraph over Groq's llama-3.3-70b-versatile
- DuckDuckGo search + calculator tool integration
- Persistent JSON memory across sessions, Streamlit UI
Legal Contract Intelligence Pipeline
Modular Python pipeline for legal contract clause extraction and summarization over the CUAD dataset.
- LLM provider abstraction spanning Groq, OpenAI, and Anthropic
- FAISS semantic search over contract clauses
- Streamlit demo app with accuracy evaluation against gold labels
Live, straight from the source.
Pulled in real time from the public GitHub API — not hand-typed numbers.
Verified credentials, listed as earned.
No certifications listed yet
This section is wired up and ready — each credential will appear here as an interactive card with a live verification link the moment it's earned.
What I can help you build.
AI Solutions
Retrieval-augmented systems, agent orchestration, and applied ML that ships past the prototype stage.
Machine Learning
Model training, evaluation, and deployment — from feature engineering to a served prediction endpoint.
Data Analysis
ETL pipelines, SQL reporting, and dashboards in Power BI or a custom web UI, built on real data.
Full-Stack Development
FastAPI or Node backends paired with React front ends — one repo, one deploy, no glue code left behind.
Automation
Multi-agent workflows and scripted pipelines that take a repetitive process off someone's plate.
Recommendations, coming as I earn them.
No testimonials yet
This carousel is built and ready to go — real quotes from mentors, professors, or clients will slot in here as they come in.
Notes on AI systems, as I write them.
Nothing published yet
This grid is ready for MDX posts — drop files into content/blog/ and they'll render here automatically (see README).
Let's talk about a role or a project.
Recruiters, hiring managers, or anyone with an interesting problem — I usually reply within a day.