About Experience Projects Skills Contact

Channabasavanna
Santosh Pawate

Python  ·  LLMs  ·  RAG Pipelines  ·  FastAPI  ·  React

Building production AI systems — from ML pipelines with SHAP explainability to RAG architectures and deep learning models. 4+ years shipping software that runs live.

4+ Years Experience
97.6% F1 Score
60% Faster Support
4 Live Systems

Who I am

Python developer with 4+ years of hands-on experience building and shipping production software. I've owned projects end to end — from REST API backends serving live e-commerce traffic to ML pipelines explaining their own predictions, RAG systems cutting support response times, and deep learning models deployed as FastAPI services with full Explainable AI outputs.

Currently completing an MSc in Artificial Intelligence at Coventry University, graduating May 2026. My dissertation achieved 97.6% weighted F1 on a COVID-19 X-ray detection task using Vision Transformer, with a complete SHAP and Grad-CAM explainability pipeline and production deployment.

Full UK right to work, no sponsorship required, available immediately. Open to London hybrid and relocation anywhere in the UK.

Location Coventry, UK — open to London hybrid
Right to Work Full UK right to work · No sponsorship required
Availability Immediately available
Education MSc Artificial Intelligence — Coventry University
BE Information Science — VTU, India

Work History

January 2021 – April 2025

AI Software Engineer

Lakshmi Silk House Consultancy — Karnataka, India

  • Built and maintained a Python FastAPI REST API backend serving a live e-commerce platform — MySQL, Docker, Linux VPS, GitHub Actions CI/CD. Four years of continuous live traffic with zero unplanned downtime.
  • Developed a production RAG pipeline using Qwen 3.5B LLM and ChromaDB — semantic retrieval over structured business documents, cutting support response time by 60%.
  • Built an ML return prediction pipeline — Random Forest (85% accuracy), full feature engineering on large retail datasets, SHAP explainability outputs, Streamlit management dashboard.
  • Built an AI Catalog Generator with Next.js 15 and React 19 — TypeScript frontend integrated with Gemini 2.5 Flash Image API, reducing catalogue asset production time by 70%.
  • Managed Linux VPS infrastructure — Docker containers, deployments, monitoring — and maintained SQL databases (MySQL, PostgreSQL, SQL Server) across multiple production systems.

August 2025 – Present

Student Ambassador

Coventry University — Coventry, UK

  • Represented Coventry University at open days, offer holder events, and campus tours — engaging with prospective students and their families to communicate course offerings and student life.
  • Delivered one-to-one guidance to prospective postgraduate students on MSc AI and technology programmes, drawing on direct experience of the course and research environment.
  • Assisted the admissions and marketing team in coordinating large-scale events, managing visitor flow, and ensuring a positive experience for hundreds of attendees per event.
  • Supported international student outreach — answered questions on visa requirements, accommodation, and campus facilities for prospective students from overseas.
  • Acted as a trusted point of contact for new and prospective students, providing peer-level insight into academic expectations, research culture, and university support services.

Selected Work

01

Production RAG Pipeline

Semantic retrieval system over structured business documents. Qwen 3.5B LLM with ChromaDB vector store, served via FastAPI, containerised with Docker for production deployment.

↓ 60% reduction in support response time

PythonFastAPIQwen 3.5BChromaDBDocker

02

COVID-19 X-Ray Detection

Vision Transformer (ViT-B/16) for COVID-19 detection from chest X-rays. Full Explainable AI pipeline — SHAP and Grad-CAM — with FastAPI inference endpoint and React frontend.

97.6% weighted F1 — MSc Dissertation

PyTorchViT-B/16SHAPGrad-CAMFastAPIReact
View on GitHub →

03

AI Catalog Generator

Next.js 15 / React 19 TypeScript frontend integrated with Gemini 2.5 Flash Image API for automated product catalogue asset generation at scale.

↓ 70% faster catalogue production

Next.js 15React 19TypeScriptGemini APITailwind
View on GitHub →

04

ReviseAI Study Bot

LLM-powered PDF ingestion pipeline — semantic chunking, automated flashcard and summary generation via Groq API (Llama 3.3 70B), with text-to-speech output.

Document → flashcards in seconds

PythonGroq APILlama 3.3 70BPyMuPDFgTTS
View on GitHub →

Technical Stack

Python

FastAPIFlaskPandasNumPyScikit-learnPyTorchSQLAlchemyPytest

AI & Machine Learning

LLM APIsRAG PipelinesChromaDBSHAPGrad-CAMVision TransformerGemini · Claude · Groq

Frontend

ReactNext.js 15TypeScriptJavaScriptTailwind CSSStreamlit

Databases

MySQLPostgreSQLSQL ServerSQLiteChromaDB

Infrastructure

DockerLinux VPSGitHub ActionsAzureVercelGit

Other

REST API DesignAgile / ScrumC# / .NETOpenAPIPostmanVS Code

Academic Background

MSc Artificial Intelligence and Human Factors

Coventry University — Coventry, UK

Dissertation: COVID-19 X-ray detection using Vision Transformer (ViT-B/16) — 97.6% weighted F1, full Explainable AI pipeline (SHAP + Grad-CAM), production FastAPI + React deployment.

Graduating May 2026

Bachelor of Engineering — Information Science & Engineering

Visvesvaraya Technological University — India

Completed

Let's work
together.

Full UK right to work. Available immediately. Open to London hybrid and relocation anywhere in the UK.