Emmanuel Johnny Koffi

The problem

Founders and investors routinely lose valuable tax breaks on startup equity because the qualifying rules are technical, depend on years of company history, and are easy to trip over without noticing. Gathering the evidence (financing rounds, balance sheets, stock agreements, operating locations) is slow, manual work for lawyers and analysts.

What I built

A multi-portal platform where investors, companies and CapGains analysts collaborate on the same data:

• Investor portal to track a portfolio of securities, see each holding's eligibility status and holding period, and get alerts when something puts a benefit at risk.
• Company portal with a step-by-step onboarding wizard that captures legal entities, parent and acquired entities, funding history, patents, office and employee locations, and financial statements.
• Guided questionnaires for QSBS (company and shareholder) and Section 1244, with answers that feed directly into the diagnostics.
• Issuer, security and shareholder diagnostics that score each criterion, flag risky answers and track every data point back to its source document.
• Generated reports, including a QSBS report and a 1244 loss report, built from dozens of reusable report sections and exportable to PDF.
• Admin and analyst tooling for coverage requests, security requests, reconciliation of schedules of investments, bulk emails and team invitations.

How it works under the hood

• Crunchbase integration (8 API calls) to autocomplete companies and investors, pull organization, people, funding-round and fund data, and pre-fill company profiles.
• AI document pipeline: uploaded stock purchase agreements are sent to a Python extraction service, and company and security diagnostics run through Langflow AI flows, with results written back into Bubble as structured answers and observations.
• Financial engine in backend workflows: income statements, balance sheets, tax-basis balance sheets, Schedule K items, capitalized R&D and amortization, financing-round tests and the 1244 transition calculation, all recalculated by chained parent/child workflows when an input changes.

By the numbers

• 21 pages
• 180+ reusable elements
• ~300 backend workflows
• 87 data types
• 130+ option sets
• 38 plugins

Bubble skills shown

Complex relational database design, privacy rules on sensitive financial data, recursive and scheduled backend workflows, API Connector at scale, AI integration, reusable-element architecture, multi-role portals, PDF generation.

The problem

Investors increasingly care where their money goes, but brokerages don't tell you how your funds and banks rank on climate. Answering that question meant exporting statements and researching every holding by hand.

What I built

• Plaid account linking for investment and depository accounts, importing accounts, holdings, securities and balances, with refresh and re-sync flows.
• CSV upload path for users whose institution isn't on Plaid, with a multi-step import that creates securities, holdings and accounts and merges duplicates.
• Climate scoring: per-holding and per-bank environmental metrics, a portfolio climate score, peer-group and benchmark comparisons, leaf ratings and a Climate Hub with contextual tips.
• Advisor matching: an advisor portal and investor matching flow that calls an external matching API and shows matched advisors with their philosophy, services, fees and booking link.
• Subscriptions with Stripe tiers, upgrade prompts and cancellation handling.
• Weekly digest emails and an admin back office for banks, sectors, funds, metrics, peer groups, tips and subscriptions.

How it works under the hood

• A numbered chain of backend workflows (create accounts, make securities, attach holdings, save balances, clean up temp records) that turns raw Plaid or CSV data into a clean portfolio without blocking the UI, with separate dev and live paths.
• Data exported to Google Cloud using a JWT-signed service token.
• Product analytics through Mixpanel.

By the numbers

• 13 pages
• 90 reusable elements
• 110 backend workflows
• 33 data types
• 32 plugins

Bubble skills shown

Plaid integration, multi-step data pipelines in backend workflows, Stripe subscriptions, external API matching, JWT and Google Cloud auth, CSV import and export, data visualization, analytics instrumentation.

The problem

An IoT connectivity business buys SIMs from several wholesale networks, each with its own portal, and resells them to customers who also need GPS trackers, invoices and support. Staff were jumping between supplier portals and spreadsheets to answer basic questions like "where is this device" or "how much data has this SIM used".

What the platform does

• Unified SIM inventory that pulls SIMs from two different wholesale providers into one list, with usage, status, network, location and provisioning history per SIM.
• Full SIM lifecycle control from inside the app: activate, cancel, swap by ICCID or MSISDN, bar and unbar, roaming settings, IMEI lock, tariff holidays, SIM groups and data caps.
• SMS centre to send single or bulk SMS to devices and read SMS history.
• Live location: SIM and cell-tower location plus GPS tracker positions on Google and Leaflet maps, with geocoding and routes.
• Commercial layer: products, supplier and customer tariffs, SIM orders, RFQs, invoices with partial payments, Stripe billing and a customer self-service portal.
• Support desk with tickets and comments, synced to the client's Monday.com support board.
• Native mobile app (14 mobile views) with a dashboard, inventory, SIM detail, SMS centre, bulk SMS and support chat.

How it works under the hood

About 100 API Connector calls across the Wireless Logic SIMPro API, a second SIM provider's API (token login and refresh), a GPS tracking server and Monday.com's GraphQL API. Nearly 50 backend workflows sync SIM lists, usage snapshots and locations on a schedule, process bulk SIM imports and handle provisioning batches. Multi-tenant organisations, partners and roles keep each customer to their own SIMs.

Size of the app I work in

• 18 pages
• 14 native mobile views
• ~50 backend workflows
• 44 data types
• 41 plugins

Bubble skills shown

Large-scale API integration, token refresh, scheduled syncs, multi-tenant privacy rules, maps and geolocation, Bubble native mobile, billing.

The problem

Teaching young people to think critically about technology needs more than a video course. Facilitators need to run classes, set and grade work and keep students engaged, and schools need oversight, all in a safe space.

What the platform does

• Four portals: student, facilitator, staff and class, each with its own navigation and permissions.
• Curriculum builder: modules, lessons, videos, readings with versions and reading levels, visuals, slide decks and a resource library with collections, folders and keyword search.
• Assignments and assessments: multiple choice, matching and open questions, polls, submissions with attachments, feedback, due-date reminders and CSV exports of results.
• Social learning: a feed with posts, reposts, likes, hashtags, @mentions, link previews and media; announcements; group and direct chat; flagging and reporting for moderation.
• Wellness and growth: meditations and guided practice with audio, playlists, "future self" goals and milestones, notes with tags.
• Notification engine for new assignments, feedback, mentions, messages and reminders by email and in-app.

How it works under the hood

165+ backend workflows drive scheduled reminders, bulk student imports, ordering of lessons and modules, search indexes and data migrations. A CSV-to-JSON pipeline lets staff bulk-upload library resources with validation. Link previews, PDF viewing with comments, audio players and an in-browser whiteboard are wired in through APIs and plugins.

Size of the app I work in

• 15 pages
• 145 reusable elements
• 165+ backend workflows
• 85 data types
• 49 plugins

Bubble skills shown

Complex role-based portals, reusable-element architecture, social features, scheduled notifications, CSV import pipelines, privacy rules for minors' data.

The problem

Architects keep their work scattered across PDFs, InDesign files and image folders, and rebuild a portfolio from scratch every time they apply for a job, pitch a client or enter a competition.

What the platform does

• Project library where each project holds its images, drawings and descriptions in one place.
• Collections to curate projects into focused sets.
• Tailored portfolio versions that can be shared with different audiences without duplicating work.
• Subscriptions with Stripe-powered paid plans and a public pricing page.

Bubble skills shown

SaaS subscription billing with Stripe, media-heavy data modelling, sharing and permissions, backend workflows.