xSU Supply Chain & Costing
Company-wide planning & cost estimation. Four real-time integrations with retries, error tracking and an early-warning system.
Software engineer with 6+ years of Python, building backend platforms, data pipelines and computer vision for Nokia and 18 clients across industries. Now building with AI.
Good systems are quiet. The real work lives in the plumbing nobody sees — the retry that catches a flaky upstream, the deduplication that keeps two sources honest, the job that runs at three in the morning and simply works.
Structure gives a system its footing. Experience shows what survives contact with production. I build data-intensive enterprise platforms — real-time ETL, integrations with error tracking baked in, document automation, delivery hardened with SSO, RBAC and row-level security.
What's changed isn't the standard — it's the tools. Computer vision has run in production under my name since 2020. Now the same discipline points at AI: classical ML, retrieval-augmented LLM apps, MLOps. Not the demo. The thing that has to keep running.
Designed the frontend and backend of a hotel and restaurant e-commerce system on Odoo. Point-of-sale, stock, purchasing and the books joined up, so what's sold at the counter flows straight to inventory and accounting.
Integrated Facebook Business Suite, Messenger, Instagram, LinkedIn, Twitter and Mailchimp into Odoo. Conversations and campaigns now sit next to the customer record, boosting customer engagement.
Implemented CyberSource (SOAP) and Paytm integrations: one enterprise card processor, one consumer wallet. Secure transactions for the platforms that needed to take money, including the taxi platform.
Built an event booking system with Django 4.1, jQuery and MySQL: listings, availability and reservations, with one source of truth for who's coming to what.
A data management web app in Python, HTML5, CSS3, Ajax and jQuery with in-page editing. It replaced spreadsheet-and-copy-paste routines and reduced manual effort.
RPA bots and API calls that pull order and database data for a cloud-kitchen operation, plus Python, VBA and Jiffy.ai scripts. Repetitive back-office steps handed to bots.
Contributed Python ML to PLC-integrated automation on production lines. Data-driven insights from the line itself lifted production-line efficiency by 25%.
Improved the image-processing stages of the ML pipelines behind real-time safety detection, owning requirement analysis, testing, debugging and code reviews.
Processed large datasets using Python algorithms and data structures to support scalable enterprise analytics: the groundwork every model depends on.
Building a classical ML project end to end while studying deep learning, LLM applications (RAG) and MLOps, bringing production habits to models.
A hierarchy selector and a CKEditor5 rich-text integration for Nokia's Odoo platforms. The editor is used every day by the document authoring team.
Refactored database schemas and API endpoints on a live Python/Django/PostgreSQL codebase, with REST integrations and automation scripts shipped in two-week sprints with security review built in.
Real-time ETL from Databricks, MSSQL, PostgreSQL, OData and REST, with retries and error tracking.
Deduplication, period aggregation and large-dataset processing with NumPy and pandas.
OpenCV object detection and OCR extraction in production; a classical ML project in progress.
Models running in real-time monitoring and PLC-integrated lines; services shipped with Docker and CI/CD.
LLM applications with retrieval (RAG), deep learning and MLOps, built on the same reliability habits.
The new engineer is hybrid. Human judgment and machine intelligence, each doing what it does best.
Six years of data plumbing taught me where AI actually earns its place: on top of clean pipelines, reliable integrations and honest data. Without those, a model is just a confident guess.
So I'm building from the ground I know. A classical ML project in progress; deep learning, retrieval-augmented LLM apps and MLOps under study — with production habits already in place: retries, observability, access control, audits.
Some things get faster: data prep, integration, iteration. Some things only become possible: vision on a production line, documents that write themselves, systems that warn you before they fail.
Mechanical engineer by degree (Anna University, 2015–2019), software engineer by curiosity. Started on the factory floor teaching cameras to spot hazards. Followed the data into automation, then backend engineering, and spent the last six years building platforms business units rely on daily. Now in Bengaluru, moving deliberately into AI application development.
CGI for Nokia · HTIC Global · Socius Innovative Global Brain · Saasvaap · VerifygnTech
Python · Django & DRF · FastAPI · Flask · Odoo v12–19 · PostgreSQL · MSSQL · MySQL · Databricks · OData · OpenCV · NumPy · pandas · Docker · CI/CD · Azure SSO · OAuth 2.0 · JavaScript · OWL · QWeb
Learning: scikit-learn · deep learning · RAG · MLOps
Python & Django Full Stack Bootcamp · Postgres Database Management · Docker Mastery with Kubernetes + Swarm · Beginner's Guide to AI · Networking Fundamentals · Operating System Fundamentals
“Most systems fail quietly. I'd rather they succeed the same way.”