What Do You Do? Frontend Engineer
Nacho joined Cleo in May 2022. We caught up with him to learn more about his role at Cleo 💙
This is some text inside of a div block with Cleo CTA
CTASigning up takes 2 minutes. Scan this QR code to send the app to your phone.
Mayank joined Cleo back in October 2023. We caught up with him to find out how he’s harnessing the latest developments in LLMs to improve Cleo’s chat.
My professional background before working with Cleo involved around five years of data science experience, and a PhD in LLM based Dialogue systems.
I’ve worked with companies like AI Enterprise, General Electric and Meredith Inc. While working at these companies, I have worked across various data science subfields such as NLP and Recommender systems, and across domains including retail and e-commerce.
Recently, I pursued a PhD. It was more niche, looking into the applications of LLMs in the dialogue systems area. I’ve been fortunate to see the LLM revolution and have utilized LLMs regularly during my PhD. During my studies I gained experience working with DialoGPT, Blenderbot, GPT, BART and BERT, to name a few. It was fascinating to observe the progress in performance with model evolution from BERT to GPT-4.
Firstly, I really liked Cleo’s mission to help people build a life beyond a paycheck. I was convinced that the way Cleo is built, from the small nudges to the friendly interactions, that a user would definitely benefit from this kind of AI assistant. Sometimes, people just need to build small habits to improve their financial health, and that’s where Cleo comes in.
I was very impressed by the overall design of Cleo's website, which felt modern and targeted at Gen-Z. I also love the Hype and Roast modes. They reflect the fact that Cleo is willing to go one step ahead of traditional financial institutions, by treating users in a friendly way and helping them to create a personal relationship with Cleo.
From my prior experience, I also knew how dialogue systems–specifically things like Cleo–work, so I was very impressed to see how smoothly the different components of Cleo’s chat work in the hands of users.
I'm working as a Data Scientist with the amazing Chat 1 squad. My job is to improve Cleo’s foundational chat capabilities, which involves new training or fine-tuning new models for chat.
Since I joined Cleo, I’ve worked on fine-tuning GPT 3.5 to achieve a similar level of performance as GPT-4, but with a reduced latency and reduced cost. I’ve also worked a little bit on the contextual intent recogniser.
I'm very excited to see OpenAI’s SORA, a video creation model.
I can' wait to see the upcoming open source models and Large Language Models which can compare to GPT but are not from OpenAI, so models like Anthropic’s Claude and Mistral Large.
I'm intrigued to see how quickly we will adopt the new models at Cleo, like we have done in the past. It will be interesting to see what more emergent ability comes out of a model, and how we can optimise the way we use AI to improve the working of Cleo for our users.
The obvious ones that we use daily at Cleo are OpenAI’s applications such as ChatGPT, SORA etc.
Some other cool apps that use AI are Google Lens and Photo Edit, which can actually edit your photo or generate a video out of an image. Stable diffusion by stability.ai has a cool close-to-reality image generation too!
Interested in joining Cleo?
Nacho joined Cleo in May 2022. We caught up with him to learn more about his role at Cleo 💙
This summer, we asked students from the University of Nottingham to spend a week at Cleo. Safe to say we learnt more from them than they did from us!
This is probably how you’d expect to round up a meeting with your manager...