Sajith Herath

Tech Enthusiast, Traveller

Follow @sajithdherath

Sri Lanka
sajithdherath@gmail.com
https://sajithdherath.github.io/

I am a Python backend engineer with 6 years of experience, dedicated to building reliable, high-performance systems that deliver results every time. My recent work focuses on integrating large language models (LLMs), AI automation, and agentic workflows-ensuring that the solutions I develop are both innovative and robust.

Through hands-on experience with scalable architectures, I have developed a strong command of the design patterns, frameworks, and tools essential for dependable software at scale. I am committed to high-quality testing and rapid validation cycles, always striving to deliver solutions that are both correct and efficient.

I enjoy tackling challenging problems at the intersection of backend engineering and artificial intelligence, and I’m always eager to learn and apply the latest technologies.

Outside of work, I’m passionate about calisthenics, traveling, and capturing the beauty of nature through photography. I value connecting with others and am always open to new ideas and collaborations.


Work Experience
  • Lead Software Engineer, PrimeSens LLC – Colombo, Sri Lanka Apr 2024 – Present
    • Designed the system architecture and led the development team to build LLM powered microservices for a Customer Relationship Management (CRM), featuring intelligent lead scoring and an agentic AI chatbot that streamlined customer interactions through deal prioritization and automated actions
    • Developed an analyzer for rent roll and T12 statements utilizing LLMs, automatically extracting tabular data from diverse PDF and spreadsheet formats and unifying inconsistent headers to accelerate real estate financial analysis
  • Senior Software Engineer, PrimeSens LLC – Colombo, Sri Lanka Aug 2021 – Apr 2024
    • Reduced AWS server costs by 89% by designing a GPU accelerated video transcoding solution (an AWS MediaLive substitute), and engineered a real-time closed captioning engine utilizing speech-to-text machine learning models
    • Developed multivariate time series forecasting models for geo-specific mortgage rates, delivering actionable insights that enhanced financial analysis and investment strategies for investors
    • Engineered a recommendation system using Collaborative Filtering and a CNN facial classifier, powering AR try-on experiences to enhance e-commerce personalization and customer satisfaction
    • Designed and deployed scalable microservices for a unified social media management platform, integrating Facebook/Instagram/X/LinkedIn APIs and managing via Kubernetes for robust performance and reliability
    • Developed high-throughput post schedulers processing complex media files and optimized data analysis across 50+ metrics, delivering key social media insights for strategy
  • Software Engineer, Self Employed Oct 2020 – Aug 2021
    • Built and scaled automated ETL pipelines using Apache Airflow for SEC Form 4 data, processing 5 million historical records and over 3,000 daily filings to power robust insider trading analytics and compliance monitoring
    • Developed and maintained web scrapers for over 8,000 companies, enriching Form 4 datasets and optimizing data processing workflows to deliver actionable trading insights and enhance investment decision making
    • Designed and implemented REST APIs (FastAPI) and microservices, including real-time video streaming solutions using AWS Chime, FFmpeg, and WebRTC, to support scalable and interactive application features
  • Software Engineer, Arimac Lanka (pvt) Limited – Colombo, Sri Lanka Jul 2019 – Sept 2020
    • Engineered multi-platform integrations for a social listening tool, expanding data extraction across various channels to enable comprehensive analysis of trends, conversations, and brand sentiment for clients
    • Integrated SERP analysis to enhance a social listening tool, enabling extraction of web search data to provide richer context and comprehensive brand insights alongside social media analysis
  • Software Engineer Intern, Fortunaglobal (pvt) Limited – Colombo, Sri Lanka Jul 2017 – Jul 2018
    • Developed fintech chatbots using fuzzy logic and sequence ML models, enabling automated handling of customer inquiries and providing secure access to essential online banking functionalities
    • Built a dashboard visualizing chatbot interactions and managed secure on-premise deployment, ensuring compliance with strict banking security standards for the conversational AI system
Education
  • University of Westminster – BEng in Software Engineering (First Class Honours) Sept 2015 - Aug 2019
Skills
  • Languages: Python, Bash, SQL
  • Libraries: FastAPI, Pandas, Celery, Numpy, Agno, LangGraph, Pytorch, Tensorflow, scikit-learn, FFmpeg, OpenCV, Selenium
  • Databases: PostgreSQL, MongoDB, DynamoDB, Redis, ArangoDB, Chroma
  • Tools & Platforms: AWS (Step Functions, EC2, Lambda, S3, Sagemaker), GCP, Docker, Kubernetes, Nginx, Apache Airflow

Let's Connect

I’m always open to collaborating on innovative AI projects or discussing the future of agentic automation.
Feel free to reach out or explore my portfolio for more!