Data Analyst | Applied Data Scientist | Python, SQL, Power BI |  Supply Chain Analytics, Operations Analytics

github

AKSHAY KANAVAJE  • Peerlist

Built an end-to-end AI compliance assistant using Python, Flask and LLM API to automate repetitive audit and compliance queries for 800+ real business clients reducing response time from hours to under 3 seconds with 24/7 availability. Engineered document intelligence using pdfplumber enables clients to upload government notices and instantly receive structured summaries with exact fines, deadlines and compliance actions eliminating manual consultant intervention for routine regulatory queries. Conducted security controls testing across the application identifying and remediating 6 critical vulnerabilities including cross-account session hijacking and 17 categories of prompt injection attacks, preventing compliance data breaches before production deployment.

AI-Driven Compliance Query Automation System

Established an automated content categorization and tagging pipeline using NLP techniques and BERT-based text representations, enabling scalable classification of unstructured text data. Improved text classification quality by applying advanced EDA, cleaning and tokenization, ensuring high-integrity model inputs and consistent tagging outputs.

Automated Content Categorization & Tagging us

- This project uses a digital circuit that can perform an alarm when it detects smoke.  The circuit comprises a Light Dependent Resistor & Light Emitting Diode. - The project is based on a major approach to control and secure homes, office, shop, markets, and universities, etc. for the fire detector main circuit design, we used (ISIS) software, and the circuit was finally implemented on a breadboard.

Automatic Fire Detector

Built a car insurance purchase prediction model using Python-based classification algorithms, enabling identification of high-propensity customer segments. Compared multiple machine learning models and evaluation metrics to balance predictive performance and business interpretability in an insurance analytics context.

Car Insurance Purchase Prediction & Risk Segm

- A Profile Copying machine is a device that can be used to copy any profile by tracing and obtaining a similar profile on a corresponding workpiece. - The shapes thus produced may be linear or any other shapes.

Design and fabrication of Profile copying mac

- The machine works on the principle of conversion of solar energy into electrical energy and then conversion of electrical energy into mechanical energy. In this project, we utilized solar energy for lifting purpose

Design and fabrication of Solar weight lifter

Built Naukri Autopilot, a Python-based browser automation bot using Playwright that automatically applies to relevant job listings on Naukri.com, reducing daily job-hunting time from 2 hours to 15 minutes. Engineered a 3-way job classification system that detects Easy Apply, company portal and question-based jobs  handling each differently and logging results across 3 CSV files with duplicate prevention across sessions. Reverse-engineered Naukri's search URL structure and implemented regex-based URL parsing to reliably extract job metadata (city, experience range), replacing fragile HTML scraping with a stable, session-persistent automation pipeline.

Naukri Autopilot : Automated Job Application 

This project implements a interactive Power BI dashboard to monitor and analyse Overall Equipment Effectiveness (OEE) across an 8-machine manufacturing plant. The solution covers equipment performance tracking, downtime classification, shift analysis, and machine-level diagnostics to support data-driven maintenance and operational decision-making. Identified plant wide OEE of 69.62% against an 85% world-class benchmark, with unplanned breakdowns accounting for 34.21% of total downtime hours and Assembly Line B emerged as the critical underperformer at 50.87% OEE with 2,083 hours lost annually, nearly 4x higher than the best performing machine. Concluded that machine reliability and shift pattern are the two primary OEE constraints, Assembly Line B and CNC Lathe 2 account for 43% of total downtime, while a 15.8 percentage point gap between Morning and Night shifts persists consistently across all machines and months.

OEE Manufacturing Performance Dashboard

Built interactive profitability dashboards using Tableau, analyzing product-, state-, and category-level sales and margin performance. Created P&L reporting views to identify revenue trends, margin drivers, and customer-level profitability insights, supporting data-driven business decisions.

Product & State-Level Profitability Analysis 

Programmed a personalized book recommendation system using collaborative filtering techniques in Python, enabling relevant content discovery based on user preferences. Deployed an end-to-end recommendation web application using Flask and Heroku, demonstrating practical model integration and delivery.

Recommendation System using Collaborative Fil

Conducted exploratory data analysis on supply chain data (sales, inventory, transportation, defect rates) using Python (Pandas, NumPy) and Plotly to understand operational patterns. Analyzed revenue by product type and SKU, and examined transportation costs and defect rates across carriers and shipping modes to identify performance differences. Evaluated inventory metrics such as stock levels, order quantities, and lead times to observe demand trends and support inventory and logistics decision-making.

Supply Chain Inventory & Logistics Optimizati

Designed a WhatsApp chat analytics tool by processing unstructured text data and performing exploratory analysis in Python, enabling automated insights into user activity, engagement, and conversation patterns. Developed a user-facing web interface to upload and analyze chat exports, demonstrating practical data ingestion, analysis, and visualization capabilities.