CV Compiler
Project highlights
About the client
About the client
Our client is an HR technology company that builds solutions to make the lives of job seekers and recruiters easier. Their initial venture, Relocate.me, is a platform that helps IT pros find a job in their desired country, whether it be Estonia, Finland or Japan. To make it possible, the platform connects candidates with relevant job opportunities.
But the founder noticed a widespread problem: lots of people were getting turned down for jobs without even getting an interview because of their resumes. So the Relocate.me team got together with candidates to fix it. Inspired by this book called "The Google Resume" by Gayle Laakmann McDowell, which is like a guide to getting hired by big tech companies like Google, Apple, and Microsoft, the team helped developers make their resumes better.
When the platform became more widely used, the team started to consider how to automate the resume enhancement process and create a tool that could improve tech resumes automatically. That's when they came up with the idea for CV Compiler. This solution uses key CV guidelines and includes a built-in glossary of technical terms to analyze a resume and suggest improvements.
To develop an ML-powered algorithm able to parse the resumes, the company addressed Flyaps. We had previously developed GlossaryTech – a Chrome extension for recruiters that highlights all the tech terms mentioned in a CV, categorizes them, and gives their definitions – and now it was the time to build a new solution. With our experience in HR tech and AI, we became a perfect match for the project.
CLIENT REQUEST
Develop an algorithm that can effectively evaluate developer resumes in different formats and offer personalized recommendations on improving them.
The goal of CV Compiler is to create a resume that makes a candidate stand out. By parsing, analyzing, and scoring tech resumes, the application maximizes users' prospects of landing their dream job.
At its core, CV Compiler is a flexible analysis tool: it effectively parses resumes, evaluating them and providing a list of tailored recommendations. As a result, the user receives unique pieces of advice that help enhance the resume and catch the attention of the top companies’ recruiters.
The solution analyzes the expertise of a candidate and, based on a built-in glossary of IT terms, evaluates which tech terms can be added to make the CV look more up-to-date and professional.
To make the app work fast and correctly, we combined the NPL algorithm with our custom machine learning model and created an API-first product. This approach allowed us to create a solution that serves both B2B and B2C models, enhancing existing parsing solutions for global enterprises.
As a result, CV Compiler became an ML-based web app that effectively processes text, evaluates resumes and highlights their weak points giving personalized recommendations to developers on how to fix them to get a job. Discover our main challenges and how we solved them to roll out an utmost tool for tech job seekers.
problem
Flyaps needed to solve the following challenges:
approach
Our decision was to build a distributed cloud-native system that integrates the tech glossary and create an ML-powered algorithm capable of discerning data in diverse formats, such as PDF documents.
solution
CV Compiler was born out of Glossary Tech, a Chrome extension that we at Flyaps had previously developed for the same client. The Glossary Tech product used natural language processing and keyword taxonomy, which we seamlessly integrated into the CV Compiler service.
To ensure optimal integration and performance, we chose a distributed cloud-native architecture, where each module operates independently. This allowed us to make changes to the system without affecting overall functionality and performance. While we focused on developing the backend, another company was responsible for the frontend.
We conducted unit testing to ensure that our distributed architecture and infrastructure were of high quality and could support the release of a reliable product.
The most popular format for CVs in the ICT field is PDF – 80 percent of all resumes parsed by CV Compiler are PDF files. The main challenge of analyzing such a format is the variety of ways in which text can be encoded.
PDF files can contain a variety of elements, such as images, tables, charts, and text boxes, and the text can be encoded incorrectly, ruining the final result. Additionally, PDF files are created using different tools, and the formatting and structure can significantly vary making it extremely difficult for an algorithm to extract the relevant information from the file.
To solve it and achieve more accurate text identification, we used libraries designed to extract text from PDF files. This allowed us to handle encoding and formatting issues and achieve better analysis. Apart from existing solutions, we used machine learning algorithms, such as natural language processing (NLP) models, trained specifically to identify and extract information from resumes.
Combining several techniques, we achieved fast and correct text extraction making the further text analysis much more accurate.
CV Compiler started as a parsing solution powered by well-known resume parsing vendor – to parse resumes, the product used third-party API, extracting information from the documents and analyzing it.
The thing is, the global external platform used by CV Compiler is oriented at the big companies that parse million-plus documents in a month. CV Compiler's needs are more specific, so working with such a big platform was quite expensive and caused certain issues – for instance, the app didn’t function at the needed speed, the resume parser had trouble understanding certain tech terms accurately, and there wasn't any proper technical support in place.
We opted to construct a bespoke neural network that substantially enhanced the accuracy and efficiency of the solution.
Flyaps created a machine learning model that fit the CV Compiler perfectly, enhancing the accuracy of identifying key sections (like a summary) and providing a consistent average response time of 1.87 second, while also providing opportunities for additional enhancements.
One of the unique features of CV Compiler is its flexible system that allows experimenting with functionality. Parsing the resumes, the algorithm does several analyses, each having its own scope of work – for example, there's a separate text analysis that checks if the verbs that describe previous work experience are in the past tense.
To make the app as flexible as possible, we developed a low-code platform that allows users to customize the solution, changing it as needed. Pre-built templates and drag-and-drop components make it easy even for non-techies to change the parsing settings, making it the right fit for numerous businesses.
API development for CV Compiler became one of the project’s milestones that allowed us to introduce API as a service business model and optimize the product for CV Compiler clients. By creating an API and offering it on a subscription basis to similar businesses, we developed a unique product that is flexible and customizable to fit the specific requirements of companies that parse resumes and need to implement custom features. One such client is Indeed, a global online job search platform that approached CV Compiler to utilize its API.
The main benefit of CV Compiler’s API is its flexibility – depending on the company’s needs, we adjust the solution, adding new features or changing the existing ones. As an example, for Indeed, we developed and added the possibility of extracting an executive summary – it was a crucial aspect for the company that wasn’t correctly supported by existing solutions.
The launch of the API enabled us to create a solution that transformed CV Compiler into a sought-after tool, successfully used by giants in the recruitment industry.
CV Compiler now functions as a two-fold solution, catering to both candidates who wish to assess their resumes and global recruitment companies, such as Indeed, who offer tailored recommendations to their users.
Serving as a B2B and B2C solution, boot camps often purchase CV Compiler to provide their students with an effective tool that maximizes their employment prospects.
With over 50,000 IT professionals using it, CV Compiler is the ultimate tool for refining a tech CV.
Result
Leveraging the capabilities of NLP and ML, CV Compiler can parse resumes and offer personalized, data-driven enhancements in less than a second.
The solution is designed to be changeable and perfectly fit custom needs – with templates and intuitive design it’s easy to adjust CV parsing as needed.
Highly customizable and effective, CV Compiler API is used by global companies like Indeed to enhance their existing solutions.
I’ve worked with internal developers before, and Flyaps has been way more motivated than they were. They challenge themselves to complete the assignment, taking time to research things fully. They definitely care about the product from a technical perspective.
Flyaps is the team that knows how to make great products – with niche expertise in AI, ML and cloud tech, we are the right fit for developing your tech solution.
Let’s discuss your project
