Sydney · Open to AI Engineer roles

33.87° S / 151.21° E

Sudipta Nath / AI Engineer

AI Engineer.Working systems. Clear evidence.

LLMs, RAG and multimodal AI, with evaluation built into the workflow. Explore my university projects in language models, clinical NLP and physical robotics.

Master of IT (AI) · Macquarie University · Completed July 2026

Have a project in mind? Work with me

+ Model behaviour is a system problem.
Inputs. Decision logic. Evaluation.

EVIDENCE, UP FRONT

Built a document-forensics LLM pipeline at Truuth: 40/40 tampered documents detected, zero false positives.

Final evaluation result; a bounded test set, not a guarantee on unseen documents.

01 Selected university projects

From the model to the whole system.

Language models. Clinical NLP. A physical robot. Three different problems, with the engineering made visible.

All projects

Industry experience / Truuth

When the model missed it,
I investigated why.

The vision encoder diluted small tampering signals. I changed the inputs and decision logic, then benchmarked the pipeline across 12 versions.

Read the engineering story
40/40modified documents detected
0false positives in final evaluation

Final evaluation result; a bounded test set, not a guarantee on unseen documents.

More explorations.

Applied ML, robotics & research

Transformer research

NeMA-LC / NeMA-Lite

Memory-augmented transformers exploring long-context reasoning.

  • Transformers
  • Neural memory
GitHub profile

Multimodal learning

In progress

StressTwinNet

Digital-twin-inspired modelling of physiological stress using multiple modalities.

  • Physiological AI
  • Digital twins
GitHub profile

Undergraduate thesis

Infectious-Disease Recognition

Transfer learning for infectious-disease recognition. Approximately 92% accuracy reported in the undergraduate thesis; dataset and split context still to be documented.

  • Transfer learning
  • Image classification
GitHub profile

02 Research & publications

Curiosity, with a method.

Research in reasoning, memory and applied machine learning.

Research project

TRACETemporal Reasoning for ADS Causal Extraction

Neurosymbolic causal reasoning for autonomous-driving crash scenarios, with LTL / SAT solver verification.

Neurosymbolic reasoningLTL / SAT verification

Supervised by Dr. Xi Zheng · Macquarie University

Research project; no publication or evaluation result is claimed.

Published work

2020

IEEE · ETCCE

CNN face-recognition attendance system

Computer vision for automated class attendance.

2023

Taylor & Francis

Predicting student satisfaction with e-learning

Research on student satisfaction in e-learning.

Descriptive paper titles. TODO: Exact citations and DOI links.

03 About & experience

An engineer who follows the evidence.

Sudipta Nath outdoors wearing headphones

I build across the full problem: data, models, application logic and evaluation. My work spans multimodal document forensics, multilingual dialogue systems and software used by a real Sydney business.

The interesting work starts when a model fails. I care about understanding why, changing the system, and testing whether the fix holds.

Download my resume

AI & language

LLMs · RAG · Prompt engineering · Multimodal inference · Evaluation pipelines · Transformers

Engineering

Python · SQL · JavaScript · REST APIs · Docker · Git / GitHub · Streamlit

Models & systems

AWS Bedrock · PyTorch · TensorFlow · scikit-learn · OpenCV · ROS2

  1. Completed June 2026

    AI Engineer Intern

    Truuth · Identity verification startup

    Built and evaluated a document-forensics pipeline using multimodal inference, forensic signals and RAG. Tracked regression behaviour across 12 versions and wrote R&D handover documentation.

  2. Completed July 2026

    Master of Information Technology

    Macquarie University · Artificial Intelligence

    Research and applied work across language models, computer vision and AI systems.

  3. 2017–2022

    BSc Computer Science & Engineering

    CUET · Bangladesh

    Undergraduate thesis on infectious-disease recognition with transfer learning.

The blog / Engineering notes

What happens
inside the build.

Practical notes on the experiments, decisions and failures behind AI systems.

Explore the blog

04 Work with me

Something your business needs built?

I also work with businesses on useful websites, AI assistants and custom AI projects.

Explore services

AI-Powered Websites

A clear, useful website built around what your customers need to do.

Start a project

Custom AI Assistants & Agents

Help customers find answers and help your team move work forward.

Start a project

AI Prototypes & Consulting

Find out what is useful and feasible before committing to a larger build.

Start a project

Real business, real platform: Stix Eatery, Miranda NSW

Let’s talk

A role, an idea,
a difficult problem.

I’m open to AI Engineer roles in Sydney and scoped AI projects.

Say hellosudipta1604129@gmail.com