Raian Khan

Founder and engineer.
I build and ship products end to end.

IT Engineer at Odgers. Building Fluxr, a scheduling optimisation platform for power-constrained data centres, and FinLit, an AI personal finance assistant, both live.

View Work Get in Touch

I build things and put them in front of people

I am a founder and engineer based in London. I run two live ventures: Fluxr, a scheduling optimisation platform for data centres in power-constrained markets, now live across five markets, and FinLit, a bank-agnostic AI personal finance assistant, live with active users.

My day job is IT Engineer at Odgers, where I manage a hybrid AWS, Azure, and M365 environment spanning EMEA, APAC, and AMER. Internal automation I built there improved ticket handling efficiency by roughly 35%.

Alongside that I build production AI systems end to end: fine-tuned LLMs, RAG pipelines, and cloud-native ML inference platforms on AWS.

Currently pursuing an MSc in Artificial Intelligence at the University of Liverpool and working towards the AWS Machine Learning Associate certification.

Things I’ve built

Illustration: a model bench where compute moves out of diesel hours and into the solar window. Not live data.

Live

Fluxr

Scheduling optimisation platform for data centres in high-sun, power-constrained markets. Decides when to run compute workloads so operators burn less diesel and cut emissions. Live across five markets: Lagos, Nairobi, Dubai, Mumbai, and San Francisco, with proxy models filling the gap where nobody else signals well, such as fixed diesel intensity for Lagos and rainfall-modulated hydro for Nairobi.

fluxr.cloud

Pulls solar forecasts, grid carbon intensity, battery state of charge, and electricity pricing into an OR-Tools CP-SAT solver, with XGBoost for solar correction. FastAPI backend on Fly.io, React dashboard on Vercel.

FastAPI, OR-Tools CP-SAT, XGBoost, React, SQLite, Fly.io

Illustration: a model bench where compute moves out of diesel hours and into the solar window. Not live data.

Illustration of compound growth over time. Not a financial projection.

Live

FinLit

Bank-agnostic AI personal finance assistant, live with active users. Ingests transaction data from any UK bank CSV export with auto-categorisation across 20+ spending categories, and includes a compound growth simulator modelling the opportunity cost of spending over 1 to 30 year horizons.

finlit.uk GitHub

Privacy-first architecture: raw transactions are processed locally and only aggregated summaries are sent to Claude Haiku for insight generation, at approx. £0.11 per user per month. Built with Next.js 14, FastAPI, PostgreSQL, and Docker.

Next.js 14, FastAPI, PostgreSQL, Claude API, Docker, Python

Illustration of compound growth over time. Not a financial projection.

Ticket Triage LLM
Personal project
Classifies free-text IT support tickets into structured, machine-readable output so triage can be automated.
Fine-tuned Qwen2.5-0.5B-Instruct using LoRA adapters on ~800 synthetic samples. Deployed as a FastAPI microservice achieving ~95% schema-valid JSON output compliance.
- PyTorch
- LoRA / PEFT
- Qwen2.5
- FastAPI
- HuggingFace
GitHub
RAG API Platform
Personal project
A retrieval-augmented generation backend ready to be dropped behind any LLM front-end, with a fully documented REST API and typed request/response contracts.
384-dim SentenceTransformer embeddings, IVFFlat-indexed pgvector store, and cosine similarity top-k retrieval hitting sub-100ms latency.
- FastAPI
- PostgreSQL
- pgvector
- SentenceTransformers
- Python
GitHub
Cloud-Native ML Inference Platform
In ProgressPersonal project
Production ML inference infrastructure designed for zero-downtime deployments, so models can ship to production continuously without manual release work.
AWS ECS Fargate behind an ALB, full infrastructure-as-code via Terraform, and a CI/CD pipeline through GitHub Actions using AWS OIDC keyless auth, with automatic container image promotion through ECR.
- AWS ECS Fargate
- Terraform
- GitHub Actions
- ECR
- ALB
GitHub

Tech stack

AI / ML

PyTorch,
LoRA / PEFT,
HuggingFace Transformers,
SentenceTransformers,
RAG Pipelines,
LLM Fine-tuning,
scikit-learn,
pgvector

Backend & Full-Stack

Python,
FastAPI,
Next.js,
PostgreSQL,
Docker,
Terraform,
GitHub Actions,
PowerShell

Cloud & Identity

AWS (ECS Fargate, ECR, ALB, IAM, EC2),
Azure / Entra ID,
M365,
Intune,
Okta

Currently studying

MSc Artificial Intelligence, University of Liverpool,
AWS Machine Learning Associate

Say hello

Open to new opportunities and collaborations.

hello@raian.uk