Software Engineer · Backend · AI/ML · Security
Building robust systems and intelligent software.
Mid-level software engineer working across the stack, specialising in production backend systems, practical AI/ML, and secure application architecture. I build reliable, maintainable software with a strong focus on performance, security, and long-term quality.
View résumé- Backend systems
- AI / ML engineering
- Cybersecurity
Portfolio focus: backend systems, AI and machine learning, cybersecurity, performance, reliability, and maintainable software.
01 / Profile
Engineering impact at a glance.
02 / Experience
Production-minded engineering experience.
Building scalable software across product engineering, infrastructure, and intelligent systems.
Coinrule (Y Combinator–backed)
Software Engineer
- Working across the stack, I integrated major perpetual futures exchanges, including Coinbase Perps and Hyperliquid, expanding the platform's automated trading capabilities.
- Developed and shipped features for the platform's mobile applications across iOS and Android.
- Improved technical SEO and site performance, achieving a 95+ Lighthouse Performance score.
- Developed agentic AI workflows to automate the creation of SEO-optimised landing pages, scaling organic content production.
Scale AI
AI Prompt Engineer
- Designed and evaluated complex programming and reasoning tasks for large language models, assessing correctness, instruction adherence, and robustness.
- Analysed model failures and provided structured feedback to improve reasoning and code-generation quality.
- Compared outputs or behaviour across models, prompts, or approaches.
03 / Projects
Selected work.
A curated set of projects showing system design, automation, AI/ML engineering, and security thinking.
Project 01 · SEO Automation
Automated SEO Content System
Built an end-to-end n8n workflow that turns keyword opportunities into published content optimised for organic search and Generative Engine Optimisation (GEO).
- Built
- An automated pipeline from keyword selection and search-intent research to content generation, on-page and GEO optimisation, and publishing directly to the website.
- Approach
- Structured content around search intent, clear topical coverage, technical SEO, and site performance—making it useful for people searching on Google and easier for AI search and chatbot systems to surface in relevant answers.
- Impact
- Reached top-five Google positions for 20+ keywords, improved click-through rate and engagement, maintained Lighthouse scores above 95, and improved the site’s visibility in AI-generated search and chatbot answers for relevant queries.
InputKeywords
→
Automaten8n
→
GenerateContent
→
PublishLive site
rankings 20+ top 5
Lighthouse 95+
workflow automated
Project 02 · AI / ML Research
LLMZip
Master’s thesis exploring how large language models can improve traditional text compression algorithms.
- Built
- A hybrid compression algorithm that leverages LLMs to model text more effectively than conventional compression methods.
- Research
- Established the lowest entropy lower bounds ever reported for multiple languages, including English, through novel cross-linguistic analysis.
- Result
- Outperformed traditional compression approaches and documented the research findings in a dedicated technical report.
InputSource text
→
ModelLLM probabilities
→
EncodeCompressed stream
modelling LLM-guided
coding entropy-based
research multiple languages
Project 03 · AI Optimisation
Adaptive Scheduler for Manufacturing
Built a proof-of-concept factory planner for CloudNC that uses CNC machine signals and AI optimisation to adapt production schedules.
- Built
- An adaptive scheduling system that converts CNC machine signals into job assignments while accounting for machine availability and operational constraints.
- Approach
- Used signal analysis and AI optimisation algorithms to re-plan work dynamically when factory events changed the operating conditions.
- Result
- Demonstrated that optimisation-driven scheduling can improve manufacturing-output planning and informed a client-specific production-ready software solution.
Live signalCNC status
Factory eventConstraints
OptimiseJob allocation
→
OutcomeAdaptive plan
inputs CNC + plant events
scheduling constraint-aware
outcome dynamic re-planning
04 / Expertise
Depth across the full lifecycle.
I combine practical delivery with analytical depth: building the product, validating its behaviour, and improving it once it is in use.
Product & Platform Engineering
I turn complex workflows into dependable software, from third-party exchange integrations and mobile features to automated publishing systems that perform in production.
- Shape clear, maintainable interfaces around evolving product requirements
- Engineer integrations with real-world failure modes and operational constraints in mind
- Use performance evidence to improve the user experience, including 95+ Lighthouse results
Applied AI & Intelligent Systems
I apply AI where it changes a decision or removes meaningful manual effort—combining rigorous evaluation with systems that can operate beyond a demo.
- Design and assess LLM tasks for correctness, instruction-following, and robustness
- Turn live data and operational constraints into decision-ready AI workflows
- Build automation with quality gates, measurable outcomes, and human value in view
Security Champion
A security champion focused on reducing application risk through secure design, adversarial thinking, and disciplined control of identity, data, and high-value transactions.
- Threat-model trust boundaries, privilege escalation paths, and abuse scenarios before code is written
- Harden APIs with strict validation, least privilege, and defensive handling of external calls
- Verify security-critical behaviour with targeted tests and make sensitive actions traceable
05 / AI & ML
Practical AI systems, not demo-only prototypes.
My AI work spans model evaluation, LLM-powered automation, and optimisation systems. I pair experimentation and research with the engineering discipline needed to turn live data and complex constraints into useful decisions.
- Benchmark outputs against clear acceptance criteria, edge cases, and known failure patterns
- Build AI workflows that automate repeatable work while retaining clear quality checks
- Use real-time signals and operational constraints to drive adaptive planning and decision-making
- Translate research ideas into measurable systems, from LLM compression to production-focused optimisation
01Signal & DataCapture · Structure · Validate
↓
02Reasoning & OptimisationModel · Constrain · Plan
↓
03Decision WorkflowGenerate · Rank · Route
↓
04EvaluationMeasure · Compare · Regress
↓
05OutcomeDeploy · Observe · Improve
06 / Cybersecurity
Security engineered from the start.
As a security champion, I treat security as a core quality attribute: understanding how a system can be misused, reducing its attack surface, and building safeguards into the paths that handle identity, value, and external input.
- Map trust boundaries, likely abuse paths, and the assets that need the strongest protection
- Design authentication and authorization around explicit permissions and least privilege
- Harden APIs and smart-contract interactions with strict validation and defensive handling of external calls
- Use targeted tests and traceable events to verify critical behaviour and support investigation
01Attack surfaceMap Assets · Entry Points · Trust Boundaries
↓
02AuthenticationVerify Identity
↓
03AuthorizationEnforce Least Privilege
↓
04Input & State ProtectionValidate · Reject · Defend
↓
05Detection & ResponseAudit · Alert · Investigate
07 / Technical Skills
Tools I use to build.
A practical toolkit for building dependable applications, intelligent workflows, and cloud-native services.
Languages
JavaScript
TypeScript
Python
C++
Go
HTML5
CSS3
SolidityFrameworks & Runtime
React
Next.js
Node.js
NestJS
FastAPIDatabases & Data
PostgreSQL
MongoDB
Redis
GraphQLTools & Platforms
Git
Docker
Kubernetes
AWS
GCP
Linux
Cloudflare08 / Education
Education & recognition.
Engineering foundations shaped by intelligent systems, low-level computing, and mathematical problem-solving.
Academic Qualifications
Imperial College London
MEng Electrical & Information Engineering · London, United Kingdom
Upper Second-Class Honours (2:1), with First-Class Honours in the Master’s year. Coursework included embedded systems, robotic manipulation, computer architecture & compilers, cryptography, distributed ledgers, deep learning, and machine learning for medical imaging.
Athens College, Hellenic-American Education Foundation
IB High School Diploma · Athens, Greece
Graduated with 44/45. Higher Level Mathematics and Physics; selected for the Mathematics and Physics Thought Club and CERN Club.
Certificates & Honours
-
Course Certificate — Cybersecurity: Threats, Defence & Risk Management
Joint programme by Hellenic National Defence General Staff (HNDGS) and National and Kapodistrian University of Athens (NKUA) · Athens, Greece
-
Course Certificate with First Degree Honours — Algorithmic Trading
Imperial College Algorithmic Trading Society · United Kingdom
-
Course Certificate — Chainlink Web3 Development
Imperial College London · United Kingdom
-
1st Award of Excellence — National Physics Competition “Aristotle”
Hellenic Association for Science and Education in Physics · Greece
-
11th Position Nationwide — National Physics Competition
Hellenic Physical Society · Greece
09 / How I Work
Principles over percentages.
Ownership
I take responsibility beyond the feature itself: understanding the problem, delivering a maintainable solution, and leaving enough clarity for the next engineer to operate and extend it with confidence.
Collaboration
I make technical trade-offs visible early and communicate in a way that helps engineers, product partners, and stakeholders move from an ambiguous idea to a shared, practical plan.
Pragmatism
I favour the smallest dependable solution that meets the real need. Complexity earns its place only when it improves reliability, safety, performance, or a clearly valuable user outcome.
Continuous Improvement
Shipping is the start of the feedback loop. I use outcomes, edge cases, and operational learning to refine performance, security, AI behaviour, and the developer experience over time.
10 / Contact
Have an interesting engineering problem?
Whether you’re exploring a complex engineering challenge, considering a collaboration, or would like to connect, I’d be pleased to hear from you.
View résumé