Software Engineer & Co-Founder

Tinhinane (Tina)
Selles

Software Engineer & Co-Founder · London

I build things at the intersection of AI and the real world — from compliance frameworks at fintech labs, to event-driven architectures at Adobe, to AI-powered startups I founded from scratch.

Tinhinane Selles
Based in London
Scroll

The
story
so far

3+ Years building
40+ Students mentored (free)
1 Women in STEM initiative
1 Startup founded

I moved to the UK at 17 — alone, with no support network, no English, and no technical background. What I did have was one unshakeable determination: to build my own venture in tech and impact as many people as possible.

Since then, I've taught myself the language, the craft, and the industry from the ground up. I broke into big tech during my bachelor's degree, moved across three cities in the UK, and am now wrapping up a Master's in Computer Science.

Along the way I mentored women in tech, co-founded two startups, and became someone who moves fluidly between engineering, research, and leadership. At the core of all of it is a deep, stubborn passion for problem solving — and a belief that where you start has nothing to do with where you end up.

Python Java React TypeScript C++ SQL AWS Docker PyTorch LLM Evaluation Multi-agent systems Kafka Django Supabase

Where I've
built things

Oct 2025 – Present Compliance Analytics

AI Student Researcher

Leading a study on prompt sensitivity and its effects on LLM reliability in regulated financial environments. Building an automated testing framework that evaluates multiple agents — OpenAI, Claude — under controlled prompt variations, generating an auditable, reusable dataset for financial use cases.

London
Jul – Oct 2024 Adobe

Software Engineering Intern

Built a Backstage plugin in React/TypeScript and integrated it into Adobe's developer portal. Used AsyncAPI to auto-generate standardised documentation across 4,000+ Kafka topics — cutting manual documentation effort by 80%. Collaborated directly with the director to unblock technical decisions across UX, Product, and Engineering.

Edinburgh
Nov 2023 – May 2024 Feral Interactive

Student Developer

Led a team of 8 developers to build an OOP-based abstraction layer in C++, replacing third-party engines while maintaining Windows and macOS compatibility. Reimplemented Atari Centipede and 5+ other games to validate the architecture. Received an Honourable Mention for technical innovation and delivery impact.

Nottingham
Sep 2025 – Present Greenwich Consultancy Club

Marketing Manager

Conducted market research on consulting trends to identify content gaps, then developed data-driven event themes that raised engagement by 15%. Closed gaps in emerging consulting topics through structured programming.

London

Things I've
shipped

001

Multi-Agent Intrusion Detection System

A multi-agent IDS where specialised agents monitor different attack types with agent-specific knowledge bases — static, dynamic, and reinforcement-updated. A confidence-based collaboration mechanism improved detection accuracy by 30% over a single-agent baseline.

scikit-learn XGBoost Neural Networks Reinforcement Learning
002

Earthquake Response Dashboard

Cleaned 83,000+ citizen-reported seismic records and built a React dashboard with three interactive views — maps, time-series, and comparison across 19 neighborhoods — to speed up damage pattern analysis during crisis response.

React D3 Recharts R
003

Bird Watching Mobile App

A mobile application for tracking bird-watching trips with photo uploads, map-based location views, and sighting records. Implemented in both Kotlin and React, with SQL and Firebase for offline-first sync and cloud upload.

Kotlin React SQL Firebase
004

Prompt Sensitivity Research Framework

An ongoing research project building an automated testing framework that generates controlled prompt variations and evaluates sensitivity across multiple LLM agents. Creating a reusable, auditable dataset for financial compliance use cases.

Python OpenAI API Claude API Research

What I'm
building

$700B Total addressable market
20–50% Return rate from shade mismatches
$45M Revenue projected by 2040
2.5% Target market capture
10–15% Conversion lift validated
BELLE — AI Beauty Layer

The decision layer beauty has been missing

BELLE is an AI-powered foundation matching engine that sits between product discovery and purchase across brands. Computer vision and personalised skin tone analysis eliminate the guesswork that costs the industry billions each year in returns and lost conversions.

  • Cross-brand AI foundation matching using computer vision and undertone analysis
  • API integration into retailer backends — self-deploy or managed buildout
  • Proprietary BeautyCloud data layer giving brands verified consumer preference insights
The Product Architecture

Three layers. One system.

01 — Consumer

The App

The face consumers see. Scan your skin, get matched to the right foundation across any brand — in seconds.

  • Face scan & skin tone analysis
  • Cross-brand product recommendations
  • Confidence scoring per match
  • Direct purchase at point of decision
02 — B2B

The API

The technical moat. Retailers and brands plug BELLE into their existing platforms — no rebuild required.

  • Self-deploy or managed integration
  • Connects to product catalogs & inventory
  • Real-time recommendations at checkout
  • Performance-based commission model
03 — Data

BeautyCloud

The long-term flywheel. Every match builds a proprietary dataset that gets more valuable with scale.

  • Verified consumer preference insights
  • Product performance across brands
  • Decision behaviour analytics
  • Licensed to brands as a SaaS layer

Beyond the
code

🌱
Build & Bloom Co.

Founder — Women in STEM Academy

Founded a free coaching academy for 40+ students on CVs, interviews, and personal branding. 98% CSAT. Ran research-backed workshops on career development and confidence building from scratch.

🎓
Community Learning, Algeria

Volunteer Tutor — Maths & French

Provided free one-to-one and small-group tutoring to 100+ students from diverse backgrounds, including autistic students and children with Down syndrome — adapting to individual learning needs.

🏅
BUCS Sports Championships

Event Volunteer

Welcomed 200+ participants and spectators, managed event check-ins, and supported smooth operations throughout one of the UK's largest university sports competitions.

Let's make
something

Open to research collaborations, engineering roles, advisory conversations, and anything that sits at the edge of what's technically interesting.

Send a message

Have a project in mind or just want to say hello?