Ketan Kumar Bagewadi | Software Developer

I’m Ketan – a Software Developer passionate about building intelligent and scalable digital solutions powered by AI and modern cloud technologies. I enjoy transforming complex ideas into fast, reliable, and high-performance applications, with experience across AI/ML systems, backend engineering, cloud-native development, and distributed architectures. I work extensively with technologies like Docker, Kubernetes, and modern DevOps workflows to build secure, scalable, and production-ready systems.

I don’t just build models — I build complete production-ready platforms. From designing AI/ML pipelines and backend services to deploying scalable applications and implementing observability with Prometheus, Grafana, and Dynatrace.

About Me

↓ Scroll to Explore

My Short Story

Ketan

Bagewadi

Driving measurable growth and engagement through thoughtful design and engineering.

Years of Experience

I am deeply interested in cybersecurity, with a strong focus on identifying phishing threats and malicious domains. Phishing remains one of the most common and dangerous attack vectors, targeting users through deceptive websites and communications. Through my work, I have actively analyzed and identified numerous phishing and malicious domains in real-time gateway and production environments, contributing to safer digital ecosystems.
To date, my analysis has led to the detection of phishing/malicious domains.

Phishing Detected

Discover my latest work and creative solutions that bring ideas to life

A selection of projects spanning AI, machine learning, backend systems, and full-stack development.

⚠ PHISHING DETECTED

URL: suspicious-bank.com

Confidence: 97.3%

Model: Deep Neural Net

Features analyzed: 48

✓ Blocked successfully

Processing 250 threads...

Live traffic: ACTIVE

AI Security · Production

Phishing & Malicious Website Detection

Python PyTorch Llama 3.1 Selenium NLP

95–97% accuracy on live traffic. Dual classification using Llama 3.1 8B with RAG + 250-thread Selenium pipeline. Also implemented Sentence-T5-XXL with cosine-similarity and zero-shot classification. Monitored with Prometheus & Grafana dashboards — 40% better visibility. Dynatrace for end-to-end observability.

Website Classifier v2

Categories: 40+

Model: Sentence-T5-XXL

Zero-shot: ENABLED

Accuracy: 94.2%

ML · NLP

Website Classification with LLMs

Sentence-T5-XXL Zero-shot Django

Deployed dual production pipeline across 40+ categories. Llama 3.1 8B with RAG for deep context understanding. Sentence-T5-XXL with cosine-similarity and zero-shot classification for lightweight inference. Served via Django REST Framework in production.

CRM Dashboard

DB Sync: 2 servers

Stack: Django + DRF

Deploy: DigitalOcean

↑ 30% faster queries

Backend · DevOps

CRM System & Business Expander

Django PostgreSQL Nginx Azure AD

Built with Django, DRF, JavaScript, Ajax, Nginx, Gunicorn, Azure AD, DigitalOcean Droplet and PostgreSQL. Synced with 2+ database servers. Achieved 30% faster read/write operations. Fully deployed on Linux with end-to-end production setup.

DeepFake Detector

Model: ResNeXt + LSTM

Dataset: FaceForensics++

Celeb-DF trained

Status: REAL ✓

AI ChatBot App

Platform: Android

API: Google PaLM-2

Backend: Firebase

Response: streaming ✓

Deep Learning · Computer Vision

DeepFake Detection & AI ChatBot

PyTorch ResNeXt LSTM Android PaLM-2 Firebase

ResNeXt + LSTM trained on FaceForensics++ and Celeb-DF datasets for deepfake video classification. Android ChatBot powered by Google PaLM-2 API with ChatGPT-like conversational flow, Firebase backend and real-time streaming responses.

Explore my journey and the technologies that define my craft.

Education

B.E. Computer Science

KLE Dr. M.S. Sheshgiri College of Engineering and Technology, Belgaum, Karnataka

2020 – June 2024

Experience

Freelance

AI/ML in Healthcare

I also work as a freelance AI/ML developer, building practical intelligent systems. I developed a speech-to-text solution using Wav2Vec2 for accurate voice-to-word conversion, where users can speak and record patient instructions, which are then converted into text and fed into Medical-Llama that analyzes the described symptoms or medical information and provides medicine-related suggestions and guidance. I also built a simple and interactive UI for seamless user interaction with the system.

Python Wav2Vec2 Medical-Llama HTML/CSS

2024

Certifications

AWS & ML Specializations

Cloud, Data Science & AI

AWS Cloud Practitioner (Amazon), Generative AI (GeeksforGeeks), Introduction to Data Science (Cisco), SQL (Great Learning), Python (Infosys Springboard).

AWS Generative AI Python SQL

Experience

Smile Security & Surveillance Pvt. Ltd.

Software Developer

Detected and flagged phishing/malicious websites with 95–97% accuracy on live traffic using Python, ML. Designed dual production classification for 40+ categories via Llama 3.1 8B + 250-thread Selenium pipeline and also with Sentence-T5-XXL. Deployed CRM with Django, Nginx, Azure AD, DigitalOcean & PostgreSQL synced across 2+ DB servers. Built network anomaly detection using DNS/NetFlow data with rule-based + Llama 3.1 8B for cyber threat detection. Implemented Prometheus for real-time ML monitoring and Grafana dashboards for accuracy, data drift & prediction trends — improving visibility by 40%. Configured Dynatrace for end-to-end observability enabling 30% faster issue detection in production.

Python AI/ML LLM Django

Mar 2025 – Present