Skip to content
View Akash-045's full-sized avatar
🎯
Focusing
🎯
Focusing

Block or report Akash-045

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.

Content in all repositories owned by your account will be closed.
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
Akash-045/README.md

👋 About Me

I started as a full-stack developer, then got more interested in what the data behind an application was actually saying than in shipping the next feature — so I made the switch, deliberately.

Today I build systems that turn messy, real-world data into decisions people can trust — and I don't stop at "it works," I check why it works before I believe it.

  • 🔭 Currently deepening my skills in Data Analytics @ Ironhack Berlin
  • 🤖 Building AI systems that validate their own output, not just generate it
  • 📊 Comfortable across the full pipeline: raw data → cleaning → modelling → dashboard → decision
  • 🎓 Google AI Professional Certificate (7 courses) & Google AI Essentials Certificate
  • 🌍 MSc International Technology Transfer Management — logistics, mobility, tech transfer
  • 💬 Ask me about: SQL, Python, Tableau, A/B testing, LLM-powered pipelines, agentic AI

🚀 Featured Projects

🤖 Autonomous BI Insight Agent An agent that investigates why a KPI moved — the way an analyst would.

  • Segments data, runs statistical tests across each cut, explains findings in plain English via the Claude API
  • Independently rediscovered a manual finding from my Vanguard A/B test project
  • Caught its own false positive from a 2-person sample — and I fixed the guardrail
  • Lesson: automation without validation is dangerous

🔗 Live Demo · 🐙 Repo

🧾 AI Invoice Processing Pipeline An LLM pipeline that doesn't just extract — it verifies.

  • Claude API extracts structured data from raw PDF invoices
  • Independent validation layer checks every result against real German invoicing rules
  • Flags inconsistencies for human review instead of trusting AI output blindly
  • Modular, GDPR-aware, deployed as a live app

🔗 Live App

🛵 Berlin Food Delivery ETA Prediction ML regression on 45,000+ real delivery records.

  • Engineered a distance × traffic feature — 7x more predictive than distance alone
  • Benchmarked 4 models, improved R² from 0.53 → 0.68
  • Reduced average prediction error to ±4.2 minutes
  • Deployed live for real-time predictions

🔗 Live App · 🐙 Repo

🍜 Germany Food Delivery Market Analysis 6,000 real reviews across Wolt, Lieferando & Uber Eats.

  • Sentiment classifier: 94% accuracy, 93% F1 on 5,793 reviews
  • Platform-specific pain points surfaced via NLP
  • Interactive Tableau dashboard with drill-down filters
  • Motivated by 5 years as a Wolt courier

🔗 Live Dashboard · 🐙 Repo


🛠️ Tech Stack

Python SQL Pandas Tableau Scikit Learn Streamlit Claude Git PostgreSQL MongoDB


📫 Let's Connect

LinkedIn Email Portfolio


"Automation without validation is dangerous." — a lesson my own AI agent taught me.

Pinned Loading

  1. bi-insight-agent bi-insight-agent Public

    Jupyter Notebook

  2. berlin-food-delivery-eta-prediction berlin-food-delivery-eta-prediction Public

    ML regression project predicting food delivery times using distance, weather, traffic, and order context. Built with real coordinates (haversine distance), feature engineering, and model comparison…

    Jupyter Notebook 1

  3. invoice-data-extractor-ai invoice-data-extractor-ai Public

    AI-powered pipeline that extracts structured data (vendor, VAT, totals) from PDF invoices using Claude, validates German VAT compliance, and exports to Excel/SQLite. Built with Python, pdfplumber &…

    Jupyter Notebook

  4. food-delivery-market-analysis-germany food-delivery-market-analysis-germany Public

    End-to-end data analysis of Germany's food delivery market (Wolt, Lieferando, Uber Eats). Covers data collection, SQL analysis, sentiment analysis, EDA, and Tableau dashboard. Built as a final proj…

    Jupyter Notebook 1

  5. customer-experience-project customer-experience-project Public

    Forked from meisnerfiona-it/customer-experience-project

    Jupyter Notebook 1

  6. Willy-Wonka-Chocolate-Factory Willy-Wonka-Chocolate-Factory Public

    Forked from mandanasoltanig/Willy-Wonka-Chocolate-Factory

    Chocolate factory management application with inventory, production, and order tracking using Python and MySQL.

    1