Latest news and updates from DSML Kazakhstan community
Stay up to date with the latest events
A long-form interview has been published with Yerzat Dulat, a pioneer of our community and an outstanding AI researcher also widely known as higgsfield.
Yerzat and his team at Higgsfield AI successfully raised $8 million in the United States for their new Diffuse platform. The mobile app lets users create video content from a text prompt from scratch, using either ready-made clips or their own photos.
Highlights from the interview:
“After school, I studied computer science for about two or three years. Later, though, I got bored and started looking for something new. That is how I discovered machine learning. In 2016, the field was still far from the hype. In fact, Kazakhstan barely had any machine-learning specialists at the time.”
“OpenAI reached out to me in 2018. After I published one of my machine-learning papers, company co-founder John Schulman emailed me and invited me to join the team. But I decided to work on my own projects.”
“Sora targets Hollywood producers, while we target marketing departments at companies that need to create large amounts of social-media content. We can help them reduce costs significantly.”
“I told investors that we are an Asian country with a strong school of mathematics inherited from the Soviet era. That appeals to Americans because they like working with Asians.”
“Within a year, AI services will be able to generate content indistinguishable from content made by people.”
The Kazakh-language Big Data Platform Meetup will take place on June 20, bringing together leading specialists in data science, engineering, and analytics.
Speakers will explain what machine learning is, why data analysis is needed, and how to approach data storage, processing, and automation. The full program will be published soon on the organizers' channel.
Registration: bigdataplatform.kz
Kazakhstani competitor Batyr Sardarbekov won a historic gold medal at the ICPC 2023 World Finals as part of the Harbour.Space University Barcelona team 🏆
Kazakhstani competitive programmers have participated in ICPC contests for 20 years, and this is the first time a participant from our country has earned a historic gold medal. Batyr Sardarbekov's victory will give a major boost to the development of this intellectual sport in Kazakhstan 🥇
🔹 Four teams from Kazakhstan also competed in the 2023 finals: Astana IT University, Kazakh-British Technical University, Nazarbayev University, and the International Information Technology University. In a tough contest, they finished ahead of 54 teams, including the National University of Singapore and the University of Melbourne.
🔹 Teams from the International Information Technology University and Astana IT University competed in the 2022 finals, finishing ahead of 79 teams, including the University of Cambridge and the University of Washington.
🌍 Kazakhstan will be the next country to host the ICPC World Finals 2024. The final will take place in Astana this September.
Congratulations to the teams—we are proud of them and wish them continued professional success!
Beeline Kazakhstan's annual Beetech Conf will take place in Almaty on April 27.
The call for conference talks is open. The program will feature timely and engaging cases designed to provide maximum value to attendees.
Applications are accepted until March 7. Follow the link and submit your proposal: Beetech 2024 Speakers.
This year will have three tracks: Big Data, Engineering, and Management (Product + Agile).
This year's Big Data focus topics are:
If you have another relevant topic, you can still submit an application and it will be considered.
Applications are open until March 7: Beetech 2024 Speakers.
Congratulations to our community member Issagali Konysbayev, who developed a unique resume-based job recommendation system using advanced NLP models.
Working with IT specialists and analysts from the Workforce Development Center, the system was adapted and deployed on enbek.kz.
The service is built with FastAPI, sentence_transformers, and PyTorch. Its core is an NLP model that converts resume and vacancy text into embedding vectors. After jobs are filtered by parameters such as region or profession, these vectors are used to calculate cosine distance between vacancy vectors and the resume query vector. The system selects the nearest vectors by cosine similarity.
A distinctive feature is that the system learns not only from current occupations, but also from the semantics of an applicant's education, skills, and previous work experience. Issagali used a checkpoint from the middle of training rather than the final model to avoid overfitting solely to current occupations.
Recommendations are fast because the system uses only a retriever model, without an additional reranker classification model. The model was trained with ContrastiveLoss. Although Issagali did not have time to test other loss functions, there remains substantial potential for further improvement.
According to a Workforce Development Center press release, the project both demonstrates important innovation in recruitment and opens new possibilities for applying NLP to real labor-market problems.
