AI / ML
RAG systems, model training, and inference that actually ships.
AI & Software Engineer
Nivakaran S.
I build intelligent, data-driven systems — end to end.
- AI / ML · Software · Data
- Open to work · Colombo, LK
Scroll
Every build starts the same way
It starts with a problem.
Not with a framework. Not with a buzzword. With a problem that's worth the work.
01 — Then, a system
I buildintelligent systemsfrom the data to the product people click.
02 — Three disciplines
Built where three fields meet.
Software Engineering
Full-stack products, end to end, built to last.
Data Engineering
Pipelines that turn raw data into something useful.
03 — Principles
Three rules I don't break.
End to end
I own the whole loop.
From the data pipeline to the button people click — no hand-offs, no gaps. One mind across the stack.
Calm under load
Precision when it counts.
Remote medical scribing and high-volume support taught me to stay exact when the pressure is highest.
Built to last
Engineered, not hacked.
Disciplined engineering over clever shortcuts. I'd rather ship something that holds than something that only demos.
04 — The stack
A connected stack, not a buzzword list.
TypeScript
Next.js
Node.js
React Native
FastAPI
Flask
Java
Kotlin
Docker
Cloud (AWS/GCP/Azure)
Python
PyTorch
TensorFlow
LangChain
LangGraph
Hugging Face
MLflow
OpenCV
CNNs
Gradio
PostgreSQL
MongoDB
Spark
Kafka
Airflow
Databricks
Pandas
NumPy
Matplotlib
Software EngineeringData ScienceComputer VisionData EngineeringData Analysis
A system is just a promise
Then it ships.
Until it's in production, handling real users, on a real Tuesday. So here's what actually shipped.
05 — The work
Eight, end to end.
Data Science01 / 08
FreeRAG
Open RAG, no API tax.
A retrieval-augmented generation system built to run without paid inference APIs — local embeddings, vector retrieval, and grounded responses over your own documents.
Data Science02 / 08
Sparrow
Intelligent parcel management with ML pipelines.
End-to-end parcel management platform with machine-learning prediction pipelines (Random Forest, XGBoost) for delivery estimation and routing — a full ML lifecycle from data to product.
Software Engineering03 / 08
EcoHarvest
AI surplus-food commerce that fights waste.
A full-stack surplus-food e-commerce platform with AI-driven recommendations, connecting vendors and buyers to reduce food waste.
Software Engineering04 / 08
MedSync
Healthcare coordination, synced.
A healthcare coordination platform streamlining clinical workflows and patient data — informed by hands-on experience in medical documentation.
Software Engineering05 / 08
LaborGuard
Workforce safety, instrumented.
A workforce management and safety monitoring application for tracking labor compliance and operational risk.
Data Analysis06 / 08
lens
See your data differently.
A data visualisation and analysis interface for exploring datasets through an interactive, modern UI.
Data Science07 / 08
NLP Book Recommendation
Embeddings that actually recommend.
A semantic book recommendation engine using Hugging Face embeddings to surface suggestions by meaning rather than keywords.
Computer Vision08 / 08
Vision CNN Classifier
Computer vision, from scratch.
A custom convolutional neural network classifying cat / dog / person images, served through an interactive Gradio interface.
06 — By the numbers
A habit of shipping, kept on the record.
- 72
- Repositories
- 5
- Competitions
- 15+
- Tech stacks
- Colombo, LK
- Based in
07 — Always sharpening
The learning never stops.
Big Data Engineering
Hadoop · Spark · Kafka pipelines
Machine Learning
Supervised & unsupervised modelling
Data Science
Analysis, statistics & visualization
Computer Vision
CNNs & image classification
Advanced ML / NLP
Embeddings & language models
Agentic AI
LangChain / LangGraph orchestration
Generative AI
LLM application development
TensorFlow
Deep learning in production
09 — The invitation
Now, let's build yours.
Open to roles and collaborations in AI/ML, software, and data engineering.