Automation Engineer / MCA & MSc Advanced Computer Science
I build systems that remove friction and actually get used — from workflow automation and document AI to edge computer vision. 100+ projects delivered. Recently completed my MSc with Distinction at Leeds Beckett University.
My instinct is to automate. When I see a repetitive process — a data pipeline, a document workflow, a manual task someone runs every day — I want to remove it. That thread has run through everything from early Fiverr scripts to the full-scale automation systems I've built more recently.
I've worked across the stack — Python, AppScript, REST APIs, GCP, Selenium, document AI — mostly on problems where success has a simple measure: does this save real time for real people?
Originally from Kerala, India. Now based in Leeds, where I recently completed my MSc in Advanced Computer Science with Distinction. What comes next is still forming — but it points at applied AI, edge systems, and software that earns its place.
Let's talkYOLOv8 nano INT8 quantised model on Raspberry Pi 5 with Flask split-inference. Edge INT8 achieved 90.6% detection at ~166ms vs cloud Float32 at 92.4% / ~117ms.
Ongoing workflow tooling — API integrations, document processing, and pipelines that cut manual overhead.
A personal ecosystem of tools, micro-automations, and utilities built around real friction in day-to-day work.
Android app built for a community organisation covering event management, daily food distribution, food ratings, donations, and real-time push notifications. Accompanied by a companion web app for event attendance tracking — organisers monitor and record participant attendance in real time. Actively used in production daily — not a portfolio demo.
100+ projects delivered across Google Workspace automation, REST API integrations, document processing, Selenium test suites, and custom business tooling. Long-term collaboration with a UK-based client on AppScript, Document AI, and GCP pipelines.
Developed and deployed an API leveraging WhatsApp Web to automate message sending at scale. Hosted online and actively used by subscribed clients for daily business communication tasks.
AI news aggregator pulling from 10+ sources. Filter by category, source, and date — then save stories into personalised digests. Fully client-side, no backend or database.
Business finance platform for loans from $50K–$50M. Covers term loans, equipment finance, and invoice factoring — with 48-hour credit decisions and a full product UI built in Next.js.
Multi-user expense reconciliation platform. Import CSVs, normalise records, and match transactions via rule-based scoring. Includes dashboard analytics and role-based access control.
12 browser-based utilities across 4 categories — text cleaning, data formatting, encoding, and calculations. Everything runs client-side with no server, cookies, or tracking.
MSc dissertation deploying a YOLOv8 nano INT8 quantised model on Raspberry Pi 5 for real-time fire and smoke detection. Core work was rigorous benchmarking across detection rate, false positive rate, inference latency, and resource utilisation under constrained hardware. Used a Flask-based split-inference architecture to keep processing at the edge without cloud round-trips.
Concluded with a cloud comparison against a Float32 pipeline — evaluating the accuracy and latency trade-offs of edge vs cloud deployment.
Open to full-time engineering roles, contract work, and interesting freelance projects. If you're building something useful and need someone who ships — I'd like to hear about it.
