Giovanni Braghieri — Product engineer
Product engineer · AI systems

I like building
products.

Lately that means AI workflows, RAG systems, agentic tools, data products, and full-stack MVPs.

I've built and sold my own products, shipped across fintech and crypto, and somewhere along the way developed a soft spot for messy problems. The method doesn't change: understand the user, shape the product, and ship something that genuinely works.

~/focus
domainfintech · crypto · edtech
buildingAI · RAG · agents
shippedproducts people use
backgroundCambridge PhD, Eng.
Engineer who ships product.

Stack

It all started with MATLAB (yes, really, lol). Then Laravel and Vue, until Vercel quietly took over the world and now I mostly speak Next.js. Python for the AI and data work, Docker and the usual suspects to hold it all together.

Laravel Vue Next.js TypeScript Python PostgreSQL Docker Git REST APIs OpenAI

Selected work

06 projects
01 AI simulation

MCC AI Interviewer

A production conversational-control system that runs structured AI case interviews: staged phases, semantic gates, action extraction, rubrics, and coach-configurable case logic.

LLM workflows AI simulation Orchestration Evaluation State machines Voice
  • ▹Multi-component orchestration engine
  • ▹Drift control along an ideal solution path
  • ▹Deterministic gates + rescue-only judge
  • ▹Declarative coach authoring, regression-tested
Read case study →
02 Data infrastructure

Fintech Data Lake & Reporting Layer

A confidential fintech data platform combining a data lake, visualization layer, dashboards, and reporting workflows across fragmented product and operational data.

Data lake SQL Dashboards Fintech Reporting Product analytics
  • ▹Data lake and reporting layer
  • ▹Product, operations, and management dashboards
  • ▹Metric definitions and reusable reporting workflows
  • ▹Worked with engineers to expose missing data
Company details and metrics cannot be shared. Read case study →
03 Crypto exchange

Tokenized Asset Exchange

A confidential order-book exchange for tokenized assets, with native tokenization and a proprietary liquidity mechanism that makes non-fungible tokens tradable as fungible units.

Order book Matching engine Tokenization Liquidity NFTs Crypto
  • ▹Standard order-book matching engine
  • ▹On-chain tokenization layer
  • ▹Proprietary liquidity mechanism for non-fungible assets
  • ▹Fungible trading with real price discovery
Company and details cannot be shared. Read case study →
04 Commercial product

MyConsultingCoach v1

A bootstrapped case-interview preparation platform supporting candidates end-to-end, from CV review to coaching, content, meeting workflows, dashboards, and admin tools.

Laravel Vue Edtech Coaching Marketplace ops Dashboards
  • ▹Paying candidates and repeat users
  • ▹Coach network and session workflows
  • ▹University and business school partnerships
  • ▹Still a live business today; v1 built on Laravel/Vue
Read case study →
05 Open-source LMS

GrindUp LMS

Open-source AI learning platform

Multi-course LMS with AI study plans, oral and written assessments, Supabase backend, Stripe billing, and teacher backoffice — evolved from a LitStudy literature prototype.

Next.js TypeScript OpenAI LMS Supabase Stripe PostgreSQL i18n
  • ▹MIT-licensed; fork and deploy your own instance
  • ▹AI oral assessment via speech-to-text + rubrics
  • ▹Multi-course backoffice with Stripe subscriptions
  • ▹Started as LitStudy literature prototype
06 RAG · Internal tools

AI Knowledge Base Assistant

A RAG-based internal Q&A assistant that retrieves answers from a curated company knowledge base, captures feedback, and surfaces knowledge gaps for continuous content improvement.

RAG Embeddings Vector search Knowledge base Feedback loops Internal tools
  • ▹Document ingestion and chunking
  • ▹Semantic retrieval over approved content
  • ▹Grounded answer generation
  • ▹Feedback loop for weak answers and gaps
Company-specific architecture and metrics cannot be shared. Read case study →