Skip to content
TSTharuka
Senior Lead AI Engineer · AI Lab, Virtusa

I engineer AI, software and data systems that solve real-world problems.

I'm Tharuka Sarathchandra, Senior Lead AI Engineer at Virtusa's AI Lab. For more than eight years I've shipped software, from Spring Boot microservices and data pipelines to machine learning, NLP and generative and agentic AI systems that I've trained, fine-tuned and taken to production. Today I work on ontologies, knowledge graphs and context engineering for agentic AI, and on compliance-first AI in regulated industries. I've also lectured at the University of Kelaniya and teach 20,000+ learners on Wizard LK.

See my work Learn with me CV
Portrait of Tharuka Sarathchandra
Software · Data · AI
8+ yrs
shipping production software
50K+
contact-center conversations analysed daily on a platform I architected
20K+
learners on Wizard LK
M.Sc.
Big Data Analytics, Distinction
  • Python
  • TensorFlow
  • NLP
  • Fine-tuning
  • Google ADK
  • Knowledge graphs
  • Ontology
  • Vertex AI
  • LangGraph
  • FastAPI
  • Elasticsearch
  • Airbyte
  • dbt
  • Airflow
  • AWS
  • Azure
  • MongoDB
  • PostgreSQL
  • Spring Boot
  • Pydantic AI
  • YOLO
  • DuckDB
01About

From a vague problem to a working solution.

The work I enjoy most starts with a fuzzy requirement: a team knows something is broken or possible, but not what to build. I sit with the people who have the problem, work out what they actually need, propose a solution across software, data and AI, and then build and deploy it with them.

That's the forward-deployed way of working, and my background covers the whole stack it needs. I trained as a software engineer, did my master's in big data, and work as an AI engineer across the full range of the field: classic machine learning such as classification and clustering, NLP models I've trained for intent classification, entity recognition and PII redaction, fine-tuning LLMs and multimodal models, the agentic and retrieval systems built on top of them, and the ontologies, knowledge graphs and context graphs that give those systems reliable context. I also know decentralized systems well, having researched and published on a blockchain-based social network built with Ethereum and Solidity.

Years of lecturing, mentoring and teaching on Wizard LK mean I can explain a design to an engineer, a manager or a customer in their own terms, which is also why I'm building Wizard Lab.

AI engineeringML, NLP, fine-tuning, agentic AI, knowledge graphs · Virtusa AI Lab
Data engineeringPipelines, warehouses, connectors · M.Sc. Big Data
Software engineeringAPIs, services, cloud, architecture · B.Sc. SE
  1. 01

    Discover

    Turn vague asks into clear problems. Workshops with users and stakeholders, mapping data sources, constraints and what success looks like.

  2. 02

    Propose

    Design the right solution across AI, data and software, with honest trade-offs on cost, risk and timeline that customers and decision makers can act on.

  3. 03

    Build & deploy

    Prototype fast, then ship to production on Google Cloud, AWS or Azure: trained and fine-tuned models, generative AI applications, ETL pipelines, APIs and integrations.

  4. 04

    Communicate & enable

    Manage customer engagements from scoping to handover, present to technical and business audiences, and train teams to own what we built.

02Experience

Where I've worked

  1. Aug 2026 – Present Colombo, Sri Lanka

    Senior Lead AI Engineer, AI Lab

    Virtusa

    Leading applied AI work in the AI Lab, from knowledge-driven agentic systems on the Google ecosystem to compliance-first AI for regulated industries.

    • Knowledge and context platform: designing ontologies, knowledge graphs and context graphs that give agentic AI systems structured, reliable context, built with Google ADK and the wider Google Cloud AI ecosystem.
    • Applying context engineering so agents retrieve, combine and reason over the right enterprise knowledge for each task.
    • Confidential project in the health and insurance domain: leading an AI-driven data conversion programme for complex, highly regulated insurance data, where compliance shapes every design decision.
    • Building agentic AI workflows and ETL pipelines for the conversion, and fine-tuning models on Google Cloud for domain-specific accuracy.
    • Ontology
    • Knowledge graphs
    • Context graphs
    • Context engineering
    • Google ADK
    • Google Cloud
    • Vertex AI
    • Agentic AI
    • Model fine-tuning
    • ETL pipelines
    • Python
  2. Jan 2023 – Jul 2026 Colombo, Sri Lanka

    Machine Learning Engineer

    Rozie Labs (Rozie AI Inc.)

    Trained and deployed ML and NLP models, and architected and led generative AI and conversation-analytics platforms for enterprise customers.

    • {"Trained and fine-tuned models across the AI stack"=>"classification and clustering, NLP models for intent classification, named entity recognition and PII redaction, and LLM and multimodal fine-tuning."}
    • Architected and led an agentic AI platform for business insight generation, with end-to-end ETL pipelines that let users ask questions in natural language and get domain-specific answers and visualizations.
    • Architected and led an enterprise RAG application combining LLMs with hybrid search and re-ranking for accurate, context-aware access to organizational knowledge.
    • Architected a conversation analytics platform processing around 50,000 contact-center conversations a day, combining agentic AI with classic ML clustering to surface patterns and trends.
    • Built a custom AI analysis harness on top of that unstructured data, so business users ask questions directly and get comprehensive reports on what is happening and why, without waiting on an analytics team.
    • Designed the data connector framework and ingestion pipelines (Airbyte, LangChain, Unstructured), and optimized Elasticsearch and MongoDB retrieval.
    • Mentored junior engineers on engineering practice and on designing and deploying cloud AI solutions.
    • Python
    • TensorFlow
    • Google ADK
    • LangChain
    • LangGraph
    • FastAPI
    • Falcon
    • Elasticsearch
    • MongoDB
    • Airbyte
    • AWS
    • Azure
  3. Jun 2021 – Dec 2022 Kelaniya, Sri Lanka

    Lecturer, Department of Software Engineering

    University of Kelaniya

    Taught databases, data science and ML, software engineering and enterprise application development.

    • Delivered lectures and practicals to undergraduate software engineering students.
    • Researched AI-driven educational technology, including a decentralized social network.
    • Supervised final-year projects applying machine learning, cloud computing and NLP.
    • Served as AWS Academy Coordinator, running cloud training and certification for students and faculty.
    • Databases
    • Data Science
    • ML
    • AWS Academy
  4. Feb 2021 – Jun 2021 Colombo, Sri Lanka

    Software Engineer

    Pearson Lanka

    Enterprise EdTech applications and an instructor recommendation system.

    • Built and maintained enterprise applications with Java, Spring Boot and MySQL.
    • Worked on an instructor recommendation system, optimizing Snowflake data pipelines and delivering insights.
    • Java
    • Spring Boot
    • MySQL
    • Snowflake
  5. Dec 2017 – Jan 2021 Colombo, Sri Lanka

    Software Engineer · Associate Software Engineer

    MVV Information Technology

    Full-stack enterprise and industrial-automation software.

    • Designed REST APIs and Spring Boot services on PostgreSQL and MySQL, plus Angular front ends and Node.js microservices.
    • Built hardware integration for industrial automation over Open Protocol.
    • Led migration of legacy application data to modern databases using ETL tooling.
    • Improved data retrieval speed by 30% through query tuning, indexing and partitioning.
    • Java
    • Spring Boot
    • Hibernate
    • Angular
    • Node.js
    • PostgreSQL
  6. Sep 2017 – Mar 2018 Colombo, Sri Lanka

    Intern Software Engineer

    Entrust & Title

    Modernized customer management systems and built financial analytics dashboards.

    • Contributed to AngularJS and .NET projects and built dashboard visualizations for financial analytics.
    • AngularJS
    • .NET
03Selected work

Projects I've architected

Platforms and products I've designed, from AI analytics at scale to computer vision.

Featured · AI & data platform

Conversation intelligence platform

Enterprise contact center

An analytics platform I architected that processes around 50,000 customer conversations a day and turns them into decisions.

  • Agentic AI combined with classic ML clustering to find patterns and emerging issues across conversations.
  • A custom-built AI harness lets business users ask questions of the unstructured data and get comprehensive reports, such as why conversations are dropping or what customers are struggling with right now.
  • Replaced manual report-building by an analytics team with self-serve, on-demand insight.
Knowledge engineering

Knowledge and context graph platform

Virtusa AI Lab

Ontologies, knowledge graphs and context graphs that give agentic AI systems structured, trustworthy context, built on Google ADK and Google Cloud.

Agentic AI · Regulated domain

AI-driven insurance data conversion

Confidential client, health & insurance

Complex, compliance-first data conversion using agentic AI workflows, ETL pipelines and models fine-tuned on Google Cloud.

Generative AI

RAG technical knowledge assistant

ZECO Energy, Australia

A RAG-powered assistant that gives electrical and energy technicians instant, context-aware technical guidance.

Data platform

Sensor analytics platform

MillenniumIT ESP

Sensor management, data monitoring and analytics platform for streams of device data.

Enterprise

Fleet, order and payment tracking

International Distillers Limited

Real-time vehicle entry/exit, route and stop tracking, usage history, BATA management and payment approvals.

Computer vision

AI identity verification

UK dating platform

Image and video verification with Amazon Rekognition to confirm users are who they say they are.

Enterprise

ERP asset transfer integration

Leading private bank, Sri Lanka

Architected and led the Asset Transfer module integration into the bank's ERP, streamlining asset workflows.

Computer vision

Sinhala number plate detection

Independent

YOLO-based detection and extraction of localized military vehicle number plates for automated identification.

Computer vision

Automated coconut counting

Independent

Real-time YOLO object detection to support agricultural productivity and resource monitoring.

Deep learning

Traditional rice species identification

M.Sc. thesis

Classifying Sri Lankan traditional rice varieties with deep learning using TensorFlow, OpenCV and Flask.

04Toolbox

Skills

Machine learning & NLP

  • Classification
  • Clustering
  • Unsupervised learning
  • Model training
  • Fine-tuning
  • NLP
  • Named entity recognition
  • Intent classification
  • PII redaction
  • TensorFlow

Generative & agentic AI

  • LLMs
  • LLM fine-tuning
  • Multimodal models
  • Agentic AI
  • Agentic workflows
  • Context engineering
  • Google ADK
  • Vertex AI
  • RAG
  • LangChain
  • LangGraph
  • Pydantic AI

Knowledge engineering

  • Ontology design
  • Knowledge graphs
  • Context graphs
  • Semantic modelling
  • OWL
  • RDF

Computer vision

  • YOLO
  • Object detection
  • OpenCV
  • Amazon Rekognition

Data engineering

  • ETL pipelines
  • Data warehousing
  • Airbyte
  • dbt
  • Apache Airflow
  • DuckDB
  • Snowflake
  • Grafana

Databases & search

  • PostgreSQL
  • MySQL
  • MongoDB Atlas
  • DynamoDB
  • Elasticsearch

Languages & frameworks

  • Python
  • Java
  • JavaScript
  • FastAPI
  • Falcon
  • Spring Boot
  • Scrapy

Blockchain & dApps

  • Blockchain
  • dApps
  • Ethereum
  • Solidity
  • Smart contracts
  • Whisper
  • Web3

Cloud & DevOps

  • Google Cloud
  • AWS
  • Azure
  • Docker
  • CloudFormation
  • GitHub Actions
  • CI/CD
Education

Learning

M.Sc. Big Data Analytics Distinction

Robert Gordon University, UK · 2021

Thesis: Identifying Sri Lankan traditional rice species using deep learning.

B.Sc. (Hons) Software Engineering

University of Kelaniya · 2019 · GPA 3.69

Certifications

  • DeepLearning.AI Data Engineering Professional Certificate 2024
  • Machine Learning in Production 2023
  • AWS Academy Educator 2022–2024

Publications

  • A Decentralized Social Network Architecture IEEE Xplore Built on decentralized technologies including Ethereum, Solidity smart contracts and Whisper messaging.
  • Identifying Sri Lankan Traditional Rice Species using Deep Learning M.Sc. thesis, Robert Gordon University · 2021 · link coming soon
05Community

Teaching in public

Wizard LK on YouTube

Since 2019 · 20,000+ subscribers

Tutorials and breakdowns on AI, ML, data science, LLMs, cloud and blockchain, in Sinhala and English.

Visit the channel →

AWS Community Builder

2022 – 2024

Workshops and online sessions on AWS, serverless and AI to help developers and students grow their cloud skills.

Lecturer & mentor

University of Kelaniya · Rozie AI · Virtusa

Taught undergraduates, supervised final-year research and mentored engineers on building and deploying AI systems.

What is artificial intelligence? Data warehousing explained What is blockchain?
Wizard Lab

Learn software, AI and data with me

Practical series on software engineering, AI and data, from someone who ships them.

  • AI Engineering
  • Data Engineering
  • AI Native Development
AI Evals | How Do We Know an AI Agent Is Working? Nov 8, 2026 AI Agents Can Write Code. But Can They Identify the Real Problem? Nov 8, 2026 AI-Native Observability | Build Software That AI Agents Can Debug Too! Nov 8, 2026 Open Wizard Lab

Let's build something.

Open to conversations about AI platforms, data engineering, consulting and teaching.

[email protected]