CareerLada - AI Career Mobility Platform

AI-Powered Career Mobility Platform for African Professionals

---

The Problem

Job searching in Africa is still largely manual, fragmented, and unoptimised. Professionals juggle multiple job boards, rebuild CVs from scratch for each application, miss scholarship opportunities they never knew existed, and have no single system for tracking where they applied or preparing for what comes next. Generic Western tools like LinkedIn Premium or Resume.io exist, but they're not priced, contextualised, or designed for the African professional market. There was no platform built specifically for this audience that combined AI-powered CV tooling, opportunity discovery, and application management in one place.

---

The Goal

Build an AI-native career hub that consolidates the entire job search lifecycle — CV building and optimisation, ATS checking, cover letter generation, interview preparation, application tracking, scholarship discovery, and salary negotiation — into a single guided workspace, purpose-built for African professionals.

---

My Role

Solo — product vision, frontend, Go backend, Python AI worker architecture, scraping pipeline, and infrastructure. The most technically complex project I have built independently.

---

What I Built

CV Studio

A real-time CV editor available without requiring a login, removing friction for first-time users. The editor renders a live preview beside the editing panel — what you see is exactly what exports. Nine templates cover the full range from minimal and ATS-safe to executive and bold. Completed CVs export to both PDF and DOCX in one click. For logged-in users, drafts auto-save across sessions. An integrated ATS quality checklist nudges users toward impact-led, metric-based bullet points that mirror target job description vocabulary — the most common reason CVs fail automated screening.

AI Agent (LangGraph + LangChain)

The intelligence layer is a Python worker service running a stateful LangGraph agent orchestrated with LangChain and powered by OpenAI. When a logged-in user triggers AI-assisted CV editing, the agent analyses the existing CV content, maps it against the target role context, and returns rewritten, scored, and improved section suggestions — functioning like a professional CV consultant rather than a basic text rewriter. LangSmith provides agent observability, tracing each reasoning step to monitor quality and catch regressions as the agent evolves.

Career Tools Suite

Beyond the CV builder, the platform exposes five additional tools: an ATS Resume Checker, a Cover Letter Generator, an Interview Prep module, a Job Application Tracker, and a Salary Negotiation Assistant — addressing the full arc of a job search from document preparation through to offer negotiation.

Scholarship Discovery Pipeline

A custom scraping pipeline built in Go using Colly as the primary scraper for static HTML scholarship boards, with Crawl4AI as the fallback for JavaScript-heavy sites that Colly can't parse. Scraped data is normalised, deduplicated, and surfaced to users as a curated, searchable scholarship feed. The same architecture is being extended for job opportunity ingestion.

---

Technical Challenges

Designing the LangGraph agent to behave reliably across varied CV structures and target roles required significant prompt engineering and stateful workflow design. LangGraph's node-based orchestration made it possible to break the CV improvement task into discrete, inspectable steps — extract, analyse, rewrite, score — rather than a single opaque LLM call. This made the agent's behaviour observable, debuggable, and improvable over time.

The scraping pipeline required handling two fundamentally different categories of web content — static HTML, where Colly excels, and JavaScript-rendered pages, which require a headless browser approach via Crawl4AI. Building a fallback routing layer that selects the right scraper per source domain, normalises the output into a consistent schema, and deduplicates across runs was a non-trivial data engineering problem for a solo build.

Running a Python AI worker service, a Go API, and a Next.js frontend as separate deployable units required careful service boundary design and inter-service communication patterns — more complex than a conventional monolithic backend.

---

Outcome

A live, functional product with the CV Studio publicly accessible without login, the full career tools suite in place for authenticated users, and the scholarship scraping pipeline operational. AI CV editing, job discovery, and additional platform features are in active development.