Skip to content
View pavankalmanoor's full-sized avatar

Block or report pavankalmanoor

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
pavankalmanoor/README.md

Hi, I'm Pavan Kumar Kalmanoor 👋

Data Analyst • Applied ML Builder • AI Systems Explorer
I build practical analytics and machine learning systems across healthcare, NLP, geospatial risk, SaaS intelligence, and agentic data workflows.

RepositoriesCurrent BuildHealthcare MLNLP Project


About Me

I’m a data and analytics professional focused on turning messy real-world data into usable systems, dashboards, and predictive products.

Right now, I’m especially interested in:

  • agentic AI for analytics
  • applied machine learning with clear business value
  • NLP and decision-support systems
  • explainable, production-minded data products

What I’m Building

Agentic Data Analyst

An LLM-powered analytics system that separates schema understanding, query planning, SQL generation, and evaluation into modular layers.

Why it matters: most “chat with your database” demos break on schema ambiguity, business metric confusion, and bad joins. This project is designed to reduce those failure modes with explicit grounding and validation.

Focus areas:

  • schema-aware prompting
  • canonical metric definitions
  • tool-using agents
  • evaluation-driven iteration
  • cost-aware design

Featured Projects

A modular AI analytics system for grounded SQL generation and layered evaluation.

Highlights

  • layer-based architecture for schema understanding, planning, generation, and evals
  • grounded data dictionary generation with canonical metrics and gotcha handling
  • cost-conscious agent workflow design

Clinical data science pipeline for predicting 30-day hospital readmission using MIMIC-III ICU data.

Highlights

  • ICD-9 comorbidity engineering and biomarker feature extraction
  • SHAP-based interpretability and fairness audit
  • Streamlit dashboard for decision support

NLP system for fan sentiment and churn-risk scoring using airline tweets.

Highlights

  • HuggingFace RoBERTa + XGBoost + MLflow
  • FastAPI serving layer
  • explainability-first workflow for classification decisions

Machine learning pipeline for hotspot detection and high-severity crash prediction using 1M US accident records.

Highlights

  • binary and multiclass severity modeling
  • threshold tuning and cross-validation
  • geospatial and temporal feature engineering

End-to-end customer intelligence project with churn prediction, LTV modeling, segmentation, retention, and dashboarding.

Regression-based analysis of psychological and behavioral factors associated with student stress.


Tech Stack

Languages & Analytics Python SQL R Pandas NumPy scikit-learn XGBoost

AI / ML Transformers Hugging Face SHAP MLflow FastAPI

Data Apps / BI Streamlit Tableau Power BI

Workflow Git GitHub Jupyter VS Code


GitHub Snapshot

Note: these cards are generated by third-party GitHub README stat services, so they can occasionally lag or fail to load.


What You’ll Find Here

  • machine learning projects tied to real decision problems
  • analytics work with practical business framing
  • end-to-end apps, not just notebooks
  • experiments in LLM systems for data workflows

Let’s Connect


Building practical AI and analytics systems that are useful, explainable, and grounded in real data.

Pinned Loading

  1. apishelf apishelf Public

    JavaScript

  2. agentic-data-analyst agentic-data-analyst Public

    Python

  3. ai-daily-digest ai-daily-digest Public

    JavaScript

  4. ai-spending-copilot ai-spending-copilot Public

    Personal finance app for tracking spending, spotting recurring charges, forecasting cash flow, and asking questions about transaction data.

    Python

  5. vaultchat vaultchat Public

    JavaScript