Skip to content
tovra

Selected work

Real products. Thoughtful engineering.

A selection of products and engineering projects spanning SaaS, AI, healthcare-related applications, and custom software.

These are products Tovra’s founder has built or contributed to — as an employee or an independent contractor — rather than Tovra client engagements. Each project states the founder’s actual role.

01Founder’s engineering work

Sparkly AI

AI Product Engineering · SaaS · Full-Stack Application

An AI-powered application-building platform that turns natural-language prompts into deployable software projects.

Work by Tovra’s founder as a backend engineer at Sparkly AI. Sparkly AI is not a Tovra client.

Founder’s contributions

  • Built the LLM orchestration layer that turns a prompt into a multi-file project.
  • Implemented streaming generation so users see progress rather than blocking on one long call.
  • Designed persistence for generated projects — file trees, revision history and per-user isolation.
Illustrative interface concept of a prompt-to-project builder — not a screenshot of Sparkly AI.

  1. Step 1

    Describe

    The user describes the site they want in plain language.

  2. Step 2

    Watch it build

    Generation streams back progressively, so the user sees files appear instead of waiting on one long request.

  3. Step 3

    Iterate

    The project is saved with its file tree and revision history, so it can be reopened and refined.

  4. Step 4

    Deploy

    A finished project is deployed live through the deployment pipeline.

Engineering challenge
An LLM is a slow, non-deterministic dependency. Generations can be partial or malformed, providers time out, and a user waiting on a multi-file project needs to see progress — not a spinner.
Solution
An orchestration layer that turns a prompt into a multi-file project and streams progress to the client, durable project persistence with per-user isolation and revision history, and retries that resume a generation rather than starting over.
  • Python
  • FastAPI
  • PostgreSQL
  • Redis
  • OpenAI API
  • WebSockets
  • Docker
  • AWS
Read the case study
02Founder’s contract engagement

Applic AI

AI SaaS · Resume Technology · Application Automation

An AI-powered resume and job-application platform designed to help users prepare tailored resumes and manage automated application workflows.

Backend built by Tovra’s founder as an independent contractor before Tovra. Frontend work was not part of this engagement.

How it works for the user

  1. 1Upload a resume. The user uploads an existing resume, which is parsed into structured data.
  2. 2Target a role. The user adds a job description to tailor against.
  3. 3Generate. A tailored resume is generated from the user’s experience and the target role.
  4. 4Apply and track. Applications run automatically in the background, each with its own visible status.
Illustrative interface concept of a resume-tailoring and application-tracking workflow — not a screenshot of Applic AI.
Engineering challenge
Auto-applying means talking to many job boards with different speeds and limits. A slow board must not stall everyone’s queue, and a partially completed run must never leave the data in an unclear state.
Solution
Queue-backed Celery workers with per-board rate limiting, so one slow upstream degrades only that board, and a data model in which every application has an explicit, queryable status.
  • Python
  • FastAPI
  • PostgreSQL
  • Celery
  • Redis
  • OpenAI API
  • Docker
  • AWS
Read the case study
03Founder’s contract engagement

SureCare AI

AI Application · Healthcare-Adjacent Technology

An AI-assisted symptom-assessment application designed to guide users through a structured symptom-related interaction.

AI integration and backend work by Tovra’s founder as an independent contractor before Tovra. Frontend work was not part of this engagement.

How it works for the user

  1. 1Start a session. The user begins a guided, conversational assessment.
  2. 2Answer structured questions. The interaction collects symptom information step by step.
  3. 3Continue safely. The session is persisted, so a refresh does not lose the conversation.
  4. 4Review a summary. The user receives a structured, readable response.
Illustrative interface concept of a guided assessment flow — not a screenshot of SureCare AI.
Engineering challenge
In a health-adjacent context, rendering unvalidated model output is not acceptable, and a slow third-party AI dependency must not leave a user stuck mid-conversation.
Solution
A prompt and response contract enforced as a schema, with invalid output rejected at the boundary; explicit timeouts, bounded retries and graceful degradation on the request path; and persisted sessions.
  • Python
  • Django
  • PostgreSQL
  • Redis
  • OpenAI API
  • Docker
  • AWS
Read the case study

Further work in fintech, e-commerce, subscription billing and real-time data is available to discuss on a call. Get in touch.

Start a project

Tell us what you’re building.

Share a few details about your product and where you are today. We’ll reply to arrange a conversation — no commitment, and no sales sequence.

Based in
Lagos, Nigeria · Working remotely with clients internationally
What happens next
Every inquiry is read by the founder. The first call is to understand the problem — not to sell you anything.

A rough range helps us suggest a realistic scope.

Links to docs or designs are welcome.

This opens your email app with your message ready to send. We use these details only to respond to your inquiry. See our privacy notice.