Experience / Resume

Sudipta Nath
AI Engineer.

AI Engineer building LLM, RAG and multimodal systems, with experience in document forensics, multilingual dialogue and full-stack delivery. Master of IT (Artificial Intelligence), Macquarie University, completed July 2026.

Experience

Completed June 2026

AI Engineer Intern · Truuth, Sydney

  • Built a Python document-forensics pipeline combining ELA heatmaps, PDF structural analysis and Qwen3-VL-235B via AWS Bedrock Converse, with a RAG component.
  • Automated false-positive/negative regression benchmarking across 12 versions. Final evaluation: 40/40 modified documents detected, zero false positives (bounded evaluation set; genuine-control count not supplied).
  • Diagnosed fine-grained signal loss from vision-encoder patch averaging; changed inputs and decision logic. Wrote R&D handover documentation.

Education

Completed July 2026

Master of Information Technology (Artificial Intelligence)

Macquarie University

2017-2022

BSc Computer Science & Engineering

CUET, Bangladesh

Selected projects

Stix Eatery Platform

Online ordering, rewards/loyalty and admin web platform for a real cafe in Miranda NSW.

View project

Smart Financial Assistant

English, Bengali and Hindi dialogue system using LLaMA 3.1 and Mistral 7B via Ollama; finite-state conversation control and a 36-trial automated evaluation pipeline.

Healthcare NLP System

8-tab Streamlit application for NER, relation extraction and clinical summarisation with fine-tuned BioBERT/ClinicalBERT.

View project

ROS2 Robot Vision & Navigation

MobileNetV3-Large classifier on a physical robot with SEARCH, ALIGN, APPROACH and ARRIVED states.

Skills

  • Python
  • SQL
  • JavaScript
  • LLMs
  • Prompt engineering
  • RAG
  • Multimodal inference
  • Evaluation pipelines
  • AWS Bedrock
  • Docker
  • Git/GitHub
  • REST APIs
  • Streamlit
  • PyTorch
  • TensorFlow
  • scikit-learn
  • Transformers
  • OpenCV
  • ROS2