Suggestions
Alejandro Fernandez
Product Manager in ML/AI
Alejandro Fernandez is a product manager and entrepreneur with extensive experience in AI/ML and startups. He is currently a Mentor at Techstars, a role he has held since August 2023.1 In this capacity, he mentors and advises pre-seed and seed-stage startups for various Techstars chapters, including Techstars New Orleans Powered by J.P. Morgan and Techstars Chicago AI.1
Professional Experience
Alejandro's current primary role is as a Product Manager for ML/AI at Square, where he has been working since May 2023.1 Prior to this, he held several notable positions:
- CEO & Co-Founder of Januso AI (June 2021 - May 2023)
- Product Manager ML/AI at Adobe (May 2022 - November 2022)
- Director of Events Product at RippleMatch (January 2021 - July 2021)
Entrepreneurial Ventures
Alejandro has demonstrated a strong entrepreneurial spirit. He co-founded Januso AI, where he built a proprietary multi-modal AI pipeline to transform English to Python using generative AI.1 This venture raised $10,000 in funding and was accepted into prestigious accelerators like UC LAUNCH and Skydeck.
He also founded Mentorade, a career accelerator for underrepresented software engineers, which helped participants land jobs at companies like PNC Bank, Dynatrace, and Booz Allen Hamilton.12
Education
Alejandro holds an MBA from the University of California, Berkeley, Haas School of Business (2021-2023).1 He also studied at the University of Melbourne and earned a Bachelor's degree in Economics & Education from Durham University, achieving First Class honors in both institutions.1
Additional Interests
Beyond his professional roles, Alejandro is involved in various side projects:
- Host of the podcast "curious," which has listeners in over 35 countries1
- Creator of "Minute Careers" on TikTok1
Alejandro describes himself as "nomadic by nature," having lived in Spain, the US, Brazil, UK, Sweden, and Australia. He is based in Miami Beach, Florida, and has a keen interest in generative AI.1
Highlights
We tried pasta, we tried dim sum, but I think yesterday's ML/AI Builder Dinner in Berlin 🇩🇪 may have proven that fish is the most loved food by the ML community.
Outstanding fish, top notch bier, and a great variety of backgrounds from folks helping identify fraud in Wikipedia posts, to folks working on optimal routing for Bolt's car sharing app, to fraud detection using passive and visual signals.
Couple interesting thoughts shared during our meal:
- LangChain can help quick prototyping, but when used at scale it becomes increasingly hard to debug which can take more time in the long-run than building your own pipeline from scratch.
- People unanimously preferred Pytorch to Tenserflow 2.
- Multiple stories of companies that had a top down mandate to "use more AI", which led to many proofs of concept that used AI, but very few actual AI products. Most didn't solve a "hair on fire" problem, if that rings a bell.
I wanted to thank the great group of attendees: Aiko Chou, Anna Melkonyan, Nurzat Rakhmanberdieva, Soso Sukhitashvili, Lukasz Dziekan, Moustafa AboulAtta, Fabian Bergmann, and Hamidreza Keshavarz.
Looking forward to continue building the ML/AI community in Europe and the US. If you're building in the space, say 👋 . Next up, we're back in Miami 🇺🇲 🏖
Nothing like sharing 100+ dim sum and exploring whether Devin AI will take our jobs or not, during the London 🇬🇧 ML/AI Builder Dinner last night.
We had a variety of backgrounds, from folks working on AI for gaming to CV for construction, and of course security. One of the prompts we discussed was "What's something unexpected you've learned while working in ML/AI?". There were three interesting takeaways to share:
- We can pitch these visionary, Jarvis-like use cases, but at the end of the day the success of any AI pipeline will heavily depend on the quality of our data, which is still a huge roadblock. We may have developed way better engines, but our fuel is still lacking. How can we increase the quality of data and the access to it?
- "Does it solve an actual problem?" still remains a key dimension to consider with any new AI product. Someone was sharing a tool they saw that helps real estate companies create AI images to help illustrate online brochures of apartments they want to rent. Images are very realistic and high quality, the issue is that since these are not actual photos they legally can only really be used as placeholders, meaning the problem of getting high quality photos is still there. How do you ground your work on problem statements vs. exciting solutions?
- Academia can sometimes incentivize overcomplicating work in an attempt to make a research paper look more groundbreaking. How do you balance the exploration of the unknown with a push for impact?
I wanted to thank the great group of attendees: Shikha Surana, Karina Dansinghani Wadhwani, Pranav Dahiya, Katerina Kolomeytseva, Hummd Ghouri, Mikel Bober-Irizar, Thien P. N., Gonzalo Barrientos, Oluwole Fagbohun, Anna Hlédiková, and Mattia Gaggi.
Looking forward to continue building the ML/AI community in Europe and the US. If you're building in the space, say 👋 . Next up is Berlin 🇩🇪
