Melbourne · Open to AI / ML & software roles

AI systems · Product engineering · Research

Rohan Prasen
Kedari.

AI ENGINEER /
SOFTWARE DEVELOPER

I build production AI that earns its place: grounded LLM applications, capable agents, and machine-learning systems measured by what they change.

Noble Park, Victoria
See selected work

40%

less manual feedback processing

5h+

saved weekly for R&I

43

agent tools implemented

14.57 dB

average reconstruction SNR

01 / Selected systems

Built beyond the demo.

Research rigor, production constraints, and interfaces people can actually use.

01

Published research · ICACCT 2026

Lossless Audio Reconstruction

GAN-based audio super-resolution that reconstructs FLAC-quality signals from MP3 input, packaged from GPU inference to a public ONNX CLI.

PyTorchFastAPIModal GPUONNX
  • + 14.57 dB avg SNR
  • + 93 reconstructions
  • + PyPI package

02

Model-agnostic coding agent

monkeyDcode

A TypeScript and Python coding agent with interchangeable LLM backends, specialist sub-agents, agentic modes, and deterministic output control.

TypeScriptPythonOpenAI Agents SDKVercel AI SDK
  • + 43 agent tools
  • + Multi-model
  • + Open source

03

Collaborative filtering research

Music Recommender

An item-based recommendation system evaluated against traditional collaborative-filtering baselines on a large-scale playlist dataset.

PythonMachine LearningData Science
  • + 2M+ playlists
  • + +12% recall
  • + Peer-reviewed DOI

02 / Experience

AI Champion
BizTX · Q1 2026

AI Engineer

Graduate Trainee

Apr — Jul 2026

BizTX · Twinings Ovaltine

  • ↳Shipped a multimodal document-to-audio platform adopted by senior leadership, saving roughly 30 minutes of reading per document.
  • ↳Automated market-signal discovery with CrewAI agents and human-in-the-loop learning, returning 5+ hours to R&I every week.
  • ↳Designed and deployed a production RAG system over internal vendor knowledge.

AI Intern

Promoted within 9 months

Jul 2025 — Apr 2026

BizTX

  • ↳Reduced manual customer-feedback processing by 40% with sentiment analysis, persona classification, cosine similarity and XGBoost.
  • ↳Built a multi-agent behaviour-cloning Q&A system for intelligent feedback collection across NPD and Sensory teams.

03 / Toolkit & trajectory

Published researcher. Production engineer. Always interested in the distance between a model working and a product working.

Languages

Python · TypeScript · JavaScript · SQL · Rust

AI / ML

RAG · Transformers · NLP · GANs · CNN/RNN · LLM APIs · MCP Agents

Build

PyTorch · FastAPI · React · LangChain · CrewAI · scikit-learn

Systems

Docker · Azure · GCP · PostgreSQL · MongoDB · MySQL

Now

Master of Artificial Intelligence

Monash University · Melbourne
2026 — 2028

Foundation

B.Tech, Artificial Intelligence

Anurag University · CGPA 7.98/10
2021 — 2025