# Aakash S S - Full Portfolio & Technical Knowledge Context

## Biographical Overview
Aakash S S is an AI & Machine Learning Engineer and Data Scientist currently based in Chennai, Tamil Nadu, India. He holds a Bachelor of Technology in Computer Science and Engineering with a specialization in Artificial Intelligence and Data Science from Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology. He has demonstrated domain expertise across Deep Learning, Computer Vision, Retrieval-Augmented Generation (RAG), and end-to-end MLOps deployment.

## Contact & Profiles
- Full Name: Aakash S S
- Email: aakash1552005@gmail.com
- Telephone: +91 88259 09003
- Portfolio URL: https://aakash1552005.github.io/
- GitHub: https://github.com/aakash1552005
- LinkedIn: https://www.linkedin.com/in/aakash-s-s-88b91a301
- Location: Chennai, Tamil Nadu, India

## Academic History
- **Undergraduate Degree**: Bachelor of Technology (B.Tech) in Computer Science and Engineering (Artificial Intelligence and Data Science)
  - Institution: Vel Tech University, Avadi, Chennai, India
  - Timeline: 2023 – Present (Expected Graduation: 2027)
  - Current Academic Standing: CGPA 7.97 / 10.0
- **Senior Secondary (12th)**: All India Senior School Certificate Examination (AISSCE), CBSE
  - School: Jain Vidyaashram, Chennai
  - Year: 2023
- **Secondary (10th)**: All India Secondary School Examination (AISSE), CBSE
  - School: Jain Vidyaashram, Chennai
  - Year: 2021

## Professional Work Experience
### 1. AI & Embedded Systems Intern | AICTE
- **Duration**: May 2026 – July 2026
- **Location**: Chennai, India
- **Key Deliverables**:
  - **Hardware – Smart Waste Segregation Bin**: Built a 5-category waste classifier (wet, dry, plastic, biomedical, recyclable) via two-phase transfer learning on MobileNetV2 (frozen-head training, then fine-tuning last 30 layers), achieving 97.76% validation accuracy. Remapped ~15K images across public datasets with class-weighted loss and augmentation. Deployed inference server integrated with ESP32-CAM and moisture sensor fallback.
  - **Software – Child Nutrition Tracking System**: Developed a cloud-based analytics platform (Next.js, Supabase, Vercel) for a government Child Development Department, applying predictive AI models to track malnutrition risk, health records, and welfare scheme delivery across child beneficiaries.

### 2. Data Scientist Intern | RINEX Technologies
- **Duration**: September 2025 – November 2025
- **Location**: Bangalore, India
- **Key Deliverables**:
  - Executed end-to-end Data Science and ML projects: data preprocessing, exploratory data analysis (EDA), feature engineering, predictive modeling, and model evaluation on business problems.
  - Developed analytical reports and interactive visualizations using Power BI and DAX to communicate actionable insights to stakeholders.

## Technical Skills & Arsenal
- **Languages**: Python (Advanced), SQL (PostgreSQL, MySQL), C++, HTML5, CSS3, JavaScript (ES6+)
- **Machine Learning & Deep Learning**: TensorFlow, Keras, PyTorch, Scikit-learn, XGBoost, LightGBM, CatBoost, OpenCV, NumPy, Pandas, SciPy, Matplotlib, Seaborn
- **NLP & Generative AI**: Hugging Face Transformers, LangChain, RAG Pipelines, Vector Embeddings (ChromaDB, FAISS), Sentence Transformers
- **MLOps & DevOps**: FastAPI, Flask, Docker, Kubernetes, Git, GitHub Actions, MLflow, Prometheus, Grafana, Jenkins CI/CD, DVC, Linux Bash
- **Data Analytics & BI**: Power BI, DAX, Excel Advanced Modeling, Statistical Hypothesis Testing, EDA (Exploratory Data Analysis)

## Detailed Project Portfolio
### Project 1: Telco Customer Churn MLOps Platform (MLOps & Production ML)
- **Problem**: ML models often fail in production due to fragmented data pipelines, inconsistent environments, manual deployments, weak lineage, and undetected drift.
- **Solution**: Engineered an end-to-end, reproducible MLOps system that automates the journey from versioned customer data to monitored production inference and model promotion.
- **Results**: Achieved 0.847 ROC-AUC, improved recall from 48% to 80% through threshold optimization, and established automated CI/CD, experiment tracking, and drift-aware retraining.
- **Stack**: Python, XGBoost, FastAPI, Docker, Kubernetes, MLflow, Prometheus, Grafana, Jenkins CI/CD, DVC.
- **Repository**: https://github.com/aakash1552005/telco-churn-mlops

### Project 2: TN Government AI Scheme Assistant (Conversational AI & RAG)
- **Problem**: Government welfare information is fragmented across 37+ scheme documents and PDFs, making it difficult for citizens to quickly find relevant benefits without hallucination.
- **Solution**: Developed a bilingual (English/Tamil) RAG conversational assistant utilizing ChromaDB embeddings and Groq LLaMA 3.3 for authoritative scheme discovery.
- **Results**: 86/86 validation tests passed, 100% pass rate on 50-query benchmark; deployed for real-time scheme discovery.
- **Stack**: FastAPI, Next.js, PostgreSQL, ChromaDB, Groq LLaMA 3.3, RAG, Railway.
- **Repository**: https://github.com/aakash1552005/TNGov-AI-Assistant

### Project 3: AI-Based Smart Garbage Segregation Bin (Computer Vision & Edge IoT)
- **Problem**: Manual waste segregation is inefficient and inconsistent; disposal points need low-cost autonomous sorting that remains functional even when cameras degrade.
- **Solution**: Built an autonomous segregation system combining camera-based real-time 5-category waste classification with sensor fallback logic.
- **Results**: Real-time 5-category waste classification with sensor-based fallback, enabling continuous autonomous segregation under both vision-enabled and degraded operating conditions.
- **Stack**: Arduino (C++), ESP32-CAM, Flask, TensorFlow.
- **Repository**: https://github.com/aakash1552005/smart-waste-segregation-bin

### Project 4: Autonomous Data Analyst (ADA) (Multi-Agent AI & Production AutoML)
- **Problem**: Enterprise data analysis is constrained by tedious manual workflows — diagnosing missingness, type coercion, masking sensitive PII, exploratory charting, baseline model tuning, compiling decks, and answering ad-hoc stakeholder questions without data hallucinations.
- **Solution**: Engineered a complete, production-grade 7-agent autonomous pipeline communicating via a validated Dataset Intelligence Object (DIO) for reversible cleaning, PII shielding, automated model competition, grounded insight synthesis, and interactive natural-language querying.
- **Results**: 100% verified numerical grounding, zero arbitrary code execution, automated PDF/PPTX report generation in <25 seconds, and 590 unit tests passing at 100%.
- **Stack**: Python, Multi-Agent System, AutoML (LightGBM, XGBoost, Random Forest, Scikit-learn), Streamlit, Flask REST API, n8n Automations, ReportLab, python-pptx.
- **Repository**: https://github.com/aakash1552005/Autonomous-Data-Analyst
- **Live Demo**: https://autonomous--data--analyst.streamlit.app/

## Verified Credentials
- **Infosys Springboard** — Exploring Data Visualization (Feb 2026)
- **RINEX (Google for Education Partner)** — Artificial Intelligence (Nov 2025)
- **RapidMiner (Altair)** — Data Engineering Master Certification (Nov 2025)
- **Deloitte** — Data Analytics Job Simulation (Sep 2025)
- **Tata Group** — Data Visualisation: Empowering Business with Effective Insights (Sep 2025)
- **Infosys Springboard** — AWS Solutions Architect: Security & Data Privacy (Apr 2025)
- **NVIDIA** — Introduction to Networking (Apr 2025)