Kasun Dulara

AI Engineer

I’m an AI Engineer at iLabs, working on AI agents, data pipelines, and backend systems. Alongside full-time work, I’ve spent the past five years freelancing on web, mobile, IoT, and machine learning projects.

Most of my day-to-day work sits between data engineering and applied ML: connecting language models to real data sources, setting up retrieval and observability, and building services that hold up in production.

Work & Projects

Xenia — AI Marketing Platform

AI Engineer · iLabs

At Xenia, I work on the agent workflows and backend for a multi-channel marketing platform. The system pulls business analytics and web data to help plan campaigns and prepare content across channels.

Worked with: LangGraph, Deep Agents, Arize Phoenix, Firecrawl, GA4, FastAPI, NetworkX, FastMCP, Redis, PostgreSQL, AWS.

Customer 360 Data Platform

Cloud of Goods · iLabs

Designed and built a streaming customer data pipeline for Cloud of Goods. It captures live user events, processes raw data across the lakehouse, and maintains unified customer profiles served through low-latency APIs.

Worked with: Delta Lake, Kafka, Debezium, Trino, dbt, Dagster, Redis, FastAPI, FastMCP, GraphQL, MySQL, PostgreSQL, AWS.

Cloud AI & FrontAI

Cloud of Goods · iLabs

Built two AI tools for an equipment rental marketplace. Cloud AI is a customer-facing assistant that answers catalog questions, suggests add-ons, and helps users with bookings. FrontAI is an internal tool for support staff that retrieves past tickets, order details, and company policies to draft replies.

Worked with: Google ADK, RAG, Milvus, Trino, Opik, Content Cloud CMS, PostgreSQL, FastAPI.

Workflow Simulation & Automation

Internship · Michelin Lanka

Modeled plant floor operations using discrete-event simulation to identify production bottlenecks and test workflow changes. Also trained XGBoost models on operational data to predict delays and built an automated pipeline to replace manual Excel reporting.

Worked with: SimPy, XGBoost, Pandas, FastAPI, PowerBI, Excel.

Independent Freelancing

5+ Years

Built custom software for clients alongside my studies and full-time roles, covering web and mobile apps, IoT prototypes, and applied machine learning projects.

Education & Research

BSc in Data Science

SLTC Research University

Studied statistics, data engineering, and machine learning. For my final-year thesis, I collected historical Sri Lankan market price data from public reports using browser automation and OCR, then evaluated custom Transformer architectures against XGBoost, LSTMs, and linear baselines.

Worked with: PyTorch, Custom Transformers, Airflow, MLflow, Gemini, XGBoost, FastAPI, Pandas, NumPy.

University Projects

A few team projects from undergrad:

  • Legal Support Agent: A RAG-based assistant for answering everyday questions about Sri Lankan law.
  • Sinhala Content Moderation: Profanity detection across Sinhala text and images using NLP and computer vision.
  • Autonomous Navigation Robot: A small ESP32 robot that processes live video streams to follow directional arrows in real time (ESP32, WebSockets, YOLOv8, Random Forest).

Certifications

Outside Work

In my spare time, I read machine learning papers and try implementing them from scratch to get a better feel for how the underlying math and algorithms work.

I put code experiments on GitHub, write occasional posts on Medium, and share short tech updates on my WhatsApp Channel. If you’d like to get in touch, feel free to send me an email.