Skip to content
← Experience

2025–2026

Hisia Space

I was the sole technical intern at an early-stage climate tech startup. In six months I shipped the company website, audited the Akili codebase and rebuilt four of its core screens, replaced Celery with Inngest to decouple long-running jobs from the request lifecycle, and ran the customer discovery that redefined the product's value proposition.

My role

I was the only technical intern at Hisia Space, through The Forge program at UM6P in Ben Guerir, from September 2025 to March 2026. I worked directly with the two co-founders, the CEO and the CTO. The CTO was based in the US. I owned four work streams: the company website, the Akili frontend rebuild with backend contributions, the Inngest integration, and customer discovery with project management.

Context

Hisia Space builds climate tools from AI, satellite data and low-cost environmental sensors. Its flagship product, Akili, is a web platform that gives farmers, NGOs and governments access to satellite imagery, weather data and pre-trained machine learning models for African climate resilience. It delivers insights over SMS, USSD and LoRa, so people in low-connectivity regions can use it.

Problem

Hisia Space had a clear vision and a team of three. It needed to get from concept to launch-ready. My brief had two parallel tracks: brand and website, and product development. On the product side that meant auditing the codebase, rebuilding core screens, contributing to the backend and proposing infrastructure improvements. The Akili codebase needed an audit and a refactor. At least three long-running processes were tied to the request lifecycle. There was no project tracking, no detailed specification and no established process. Product direction still needed validating with the people Akili was meant to serve.

What I built

Company website

The startup had no public presence, so I shipped one. I created a Figma design system with a color palette, typography and reusable components, designed a new logo and wrote brand guidelines. I built the site in React, with Three.js for 3D assets and Framer Motion for animation, and optimized it for performance, responsiveness and SEO. It is deployed to Vercel with a CI/CD pipeline, so every change ships automatically, and it is live. The design system and guidelines gave the team a base to build every later page and product screen on.

Akili platform

I audited the full codebase: frontend architecture, API contracts and technical gaps. The findings drove what I rebuilt. I rebuilt four core screens in Next.js with a component-based architecture, which made the UI consistent, responsive and reusable, and put the product on a launch-ready footing. On the backend I made targeted contributions to FastAPI endpoints and business logic, and worked on PostgreSQL data modeling and query optimization.

Background jobs with Inngest

At least three long-running processes were tied to the request lifecycle, which hurt reliability. I proposed moving from Celery to Inngest and documented the trade-offs. I built a proof of concept with three event-driven pipelines, and after the co-founders approved I integrated Inngest for background job orchestration. That decoupled three or more long-running processes from the request lifecycle and improved platform reliability.

Customer discovery and project management

Before more was built, I mapped and contacted UM6P researchers, agricultural NGOs and government agencies in Morocco. I ran 10+ structured interviews of 15 to 30 minutes about their workflows, pain points and decision-making, and wrote up the findings as reports. They redefined the product's core value proposition. I also researched soil, hydrology, satellite and agronomy datasets for the AI and LLM layer. To track the work, I built a GitHub Projects board from scratch, with a custom workflow and an issue for every task across the website and platform tracks. The CTO followed progress, reviewed work and validated fixes asynchronously, without constant meetings. I documented decisions, technical approaches and blockers.

Key decisions and trade-offs

  • Replace Celery with Inngest for background jobs.

    Trade-off: I did not change the stack on opinion. I documented the trade-offs of both tools and built a proof of concept with three event-driven pipelines to validate the approach. The co-founders approved before I integrated it.

  • Run customer discovery before building further.

    Trade-off: I ran it alongside the website and platform work, which added to the context-switching. The interviews showed what users needed instead of what we assumed, and they redefined the core value proposition.

Challenges

  • Parallel work streams. The website, the platform, the backend and the research all ran at once, and I switched context constantly. I worked in focused blocks and kept clear documentation to reduce the overhead.
  • No specifications. There were no detailed specs and no established processes. I defined scope, made architectural decisions and validated them with little guidance. I documented each decision so the team stayed aligned.
  • Shifting priorities. Priorities changed often as the company reacted to market feedback. I stayed flexible and told the CEO and the CTO early how a change affected timelines.

Outcome

The startup went from no public presence to a live website with a design system, a new logo and brand guidelines. Akili was launch-ready by the end of the internship, with a codebase audit, four rebuilt core screens and background jobs decoupled from the request lifecycle through Inngest. The discovery interviews redefined the core value proposition before more was built. The GitHub Projects board let the CTO follow and validate work without constant synchronous meetings.

What I learned

Building the right thing started with talking to users, not with the codebase. Discovery and project tracking were not in my brief, but the company came to depend on them. Working across product, design and engineering made me more useful at this stage than depth in one area would have.

Stack

  • React
  • Next.js
  • TypeScript
  • Three.js
  • Framer Motion
  • Tailwind CSS
  • FastAPI
  • PostgreSQL
  • Inngest
  • Figma
  • Vercel
  • GitHub
  • Docker