About Me
20+ years in engineering, the last few leading ML and data engineering for Grab’s Trust, Identity & Safety team in Singapore. I build the fraud detection models and data pipelines that watch transactions across Grab’s platform. Lately that work has shifted toward LLM applications: device checks, tamper detection, automated fraud reporting, and most of my own coding now runs through agents like Claude Code. Before Grab I built banking, e-commerce and aerospace platforms, which is where the systems design habits come from.
Core Competencies:
Engineering Leadership
- Led teams of up to 10 engineers across fraud detection, data platform and portal projects, from design through production support
- Pushed LLM-assisted development and agentic workflows into the team’s daily practice, starting from my own experiments
- Data Architect for enterprise-scale data lakes and warehouses; Solution Architect on AWS cloud native systems
- Take ambiguous problems and own them until they ship
AI Engineering & LLM Tooling
- Prompt engineering and RAG systems, running in production
- ML pipelines built with MLOps practices, from training through monitoring
- Autonomous agents for business process automation, plus agentic workflow experiments with OpenAI and Hermes-family frameworks
- Daily driver tools: Claude Code, Cursor, Codex
Domain Expertise
- Fraud Detection & Risk Management - real-time ML systems watching millions of transactions
- AI-powered products - LLM applications, agents and internal AI tooling in production
- Banking Systems (Merchant & Customer Portals)
- Order Management & E-commerce
- Data Center Operations
- REST-based Middleware & Microservices
Experience
Grab - the leading super app in Southeast Asia - combines transport, food delivery, logistics, financial services, and much more in a single platform and its mobile application in Cambodia, Indonesia, Malaysia, Myanmar, the Philippines, Singapore, Thailand, and Vietnam. The company is headquartered in Singapore
The TIS (Trust, Identity & Safety) department acts as guardians of all users on Grab. It closely monitors every event such as account login, booking and transactions on the platform for over 8 major business lines across Grab, making data driven decisions in milliseconds. The association with current organization helps to address the cutting edge industrial fraud detection problem to solve within an year of association.
Modules Lead and contributed:
- Work on problems that run the gamut from data science to deploying and monitoring models in production.
- Build and deploy machine learning models for fraud, risk and safety. Scam rings across Southeast Asia keep getting more organized and automated, which has made fraud detection one of Grab’s most closely watched business lines.
- Built 10+ internal skill plugins that turn raw case data into Modus Operandi (MO) reports, cutting down the manual work investigators used to do piecing together how a scam actually played out.
- Designed and maintain 10+ LLM-based applications in production - device information analysis, food preference inference, and tamper detection that catches cloned apps, auto-clickers and other scripted abuse before it reaches real users.
- Analyze data and define metrics for feature evaluation and model performance.
- Design and implement robust data pipelines.
- Identify and build new approaches and methods for machine learning as we grow.
- Do most day-to-day coding with Claude Code, Cursor and Codex.
- Prototyped agentic workflows with OpenAI and Hermes-family agent frameworks, including a Slack bot that pulls fraud signals on demand.
- Framework and technical: Java 11, Scala, Flink, Elasticsearch, Kafka Streaming, Terragrunt, AWS EC2, S3 and DynamoDB.
Equinix
Senior Staff Engineer (Lead Engineer)
December 2015 - September 2021
customerportal.equinix.com
Equinix's single global platform helped digital leaders stay out front for more than 20 years. Interconnect your digital infrastructure to scale with agility, leveraging leading enterprises and service providers in ecosystems that multiply value for all. At Equinix, we see every day as a fresh opportunity to help you fast track your digital advantage.
Equinix’s Colocations platform provides interconnections between digital enterprises. The customer can avail the colocations services offered and provides all the aftermarket facilities like billing in a single unified platform. Customers can utilize multiple channels, web, mobile, and Partner API for direct Integration. The organization has given me the most challenging data center industrial problem to solve with 6+ years association.
Modules Lead and contributed:
- Enterprise Search Engine - Search is a permission drive search engine; this allows the user to multilingual search queries to find relevant information. Relevancy and racking are designed based on business domain fields to provide a seamless user experience.
- Unified Notifications service - Unified notification message services with services like firebase, MS teams, Azure queue, SMS, and email.
- Interconnection LOA management - The LOA process helps an approval process that allows establishing connection between two customers.
- Machine learnings projects
- Smart Discovery - Discovering the user actions based on users’ attachment or description. Based on attachment content or request descriptions, the ML model will recommend order types and pre-populate the necessary details.
- Order Recommendation - This ML models recommend that the user place the connected orders based on frequent action. ML model allows to reduce the number of clicks and predicts the dependent orders.
FIS stays ahead of how the world is evolving to power businesses, across merchants, banking and capital markets, to outpace today’s fast-changing competitive landscape and help our clients run, grow and achieve more for their business.
The organization needs banking solutions online to facilitate the buyer and seller to check the balance and make payments. I was the lead engineer to provide generic solutions for merchants and customers to view and manage any transaction life cycle. Portal will help the FIS Customer get them out of the box solutions with a white labeling option. The organization has given me multiple opportunities to mature my industrial banking experience in a short period of 1.6 years.
Modules Lead and contributed:
- Merchant Portal - Facilitates mPOS device association with seller. This portal provides seller to manage successful transactions and refunds. The frameworks used in this project are Angular JS as single page applications backed by middleware technology Spring 4.x, Java 8 and Nginx
- Prepaid Customer Portal (Mobile wallet)- Facilitates mPOS device association with seller. This portal provides seller to manage successful transactions and refunds. The frameworks used in this project are Angular JS as single page applications backed by middleware technology Spring 4.x, Java 8 and Nginx
- OTP platform Service - One time password solutions used as 2 factor authentications for any payment authorizations or authorizing user for verifications. The frameworks use in this project are Apache Camel, Scala, SMS Gateway services.
- POC:
- Aadhaar based KYC approval for customer - Creating back account based on aadhaar mobile number and finger print solutions to create a back account. The frameworks use in this project are Angular JS, Apache Cordova, finger print scanner, android.
HTSL (Honeywell Technology Solutions Lab)
Technical Lead Engineer
July 2006 - April 2014
aerospace.honeywell.com
Honeywell Technology Solutions Lab Private Limited is a computer software company
Portal has the ability to airplane product orders and manages the warranty for Honeywell aerospace-related offerings. Portal facilitates engineers to download engine technical documentation based. Customers will have the option to make payment for the rented airplane engines or services offerings. During 7.6 years of the journey, organizations have given me multiple opportunities to mature aerospace industrial experience. I played various roles from time to time for 16 months as an on-site coordinator, five years as a technical lead with a team size of 10 members, two years as an engineer, and six months as an intern trainee. During my organizations’ journey, I was an active member in evaluating technology and innovative solutions.
Modules Lead and contributed:
- Aerospace Portal - A leading web portal for Honeywell Aerospace customer and Customer support representatives to place orders, collate information on the order, download the technical document, and search for field service engineer and support locator. The frameworks used in this project are Websphere Portal 7, Sprint Portlets and Generic portlets.
- SHARED Portal Migration - Enable Honeywell’s airline customers to access value-added aftermarket service for their MRO requirements through migration of SHARED system functionality to the MyAerospace portal. Solutions is based on Spring 3.x, MyBatis, Ajax and Websphere portal server.
- Plymouth Migration - Reselling portal enabling customers & employees to search & browse through the catalog, add to cart and process with placing an order. The frameworks used in this project are SAP ECC eSelling, XCM, NWDS.
- ATR Catalog - Enable internal Honeywell’s sales rep to access ATR catalog form any mobile device. This allows configuring and downloading the catalog online and persists them on the devices. Especially when they are in the Aero show conference
- Other Projects
- HSCE - Honeywell Health and Safety system Portal
- Teardown Management system - Barcode generation, for comparing LCD productions for Honeywell Patterns
- HonBoard - Online whiteboard solutions like “Miro”
- Catalogic - IPad Applications for Aerospace Catalog for sales and marketing.
- Integrated Supply chain - SAP Enterprise Portal Solution based for Supply chain management.
- Single Singe On Integration - using saml defining authorizations and authentications between multiple systems
Education
The University of Texas, Executions Educations
Post Graduate certificate Program - Artificial intelligence and Machine learning
2019 - 2020
eportfolio.greatlearning.in/kamesh
The University of Texas at Austin, shortened to UT Austin, UT, or Texas, is a public research university in Austin, Texas and the flagship institution of the University of Texas System.
I learn to create and manage machine learning models during the modern way of online education—the basics skill like cleansing the data for model prediction, PCA, and Feature engineering. I learned to organize the data and understand the business use case to predict the models.
JJ College of engineering (Affiliated to Anna University)
BE Computer Science
2003 - 2006
JJ Colleger of Engineering affiliated to Anna University
During my time at College I learnt most of my key skills that have I have taken through my career such as teamwork and working to tight deadlines. I thoroughly enjoyed my time as university and learnt a lot about a healthy work life balance.
I spent a lot of my free time as a committee member of the College Developer Community taking on roles such as arranging hacking sessions for juniors Hackathon Organizer and Cultural Event Vice-President.
Department of Technical Educations
Diploma in Computer Technology
2000 - 2003
KAR Polytechnic college, Ambur, Tamilnadu, India