Meet ABI AI in the real world, aigogo & Deep Listening AI

The MLAI Meetup is a community for AI researchers and professionals which hosts monthly talks on exciting research. Our format is:

• 6:00 - 6:20: Socializing
• 6:20 - 6:40: Announcements and AI news
• 6:40 - 7:40: Talk(s) and Q&A
• 7:40 - 8:00 Networking
• 8:00: Head to the nearest pub for dinner

THREE amazing talks tonight:

1. Chaehan So: “Deep Listening AI: A Psychologist's Blueprint for Building a Virtual Friend”
2. Dushan Karovich-Wynne: “aigogo: Packaging AI Agents for Production”
3. Adam "Spidey" Purdie & Rohan Nowell: "Meet ABI AI in the real world"

Talk descriptions and speaker bios:

Adam "Spidey" Purdie & Rohan Nowell: "Meet ABI AI in the real world"

Synopsis:
Andromeda's purpose is to build technology that strengthens connections and preserves purpose in people’s lives. This purpose shapes what we build and how we build it. The soul of the product matters as much as its capabilities, and our values are encoded through everyday decisions.

Abi is our friendly humanoid robot who is built for this purpose. Where she most strongly differs from a LLM chatbot is in embodiment, she shares a physical environment with you. This comes with a whole new range of challenges, she doesn’t just need to know what to say to hold a conversation, she now also needs to control her body language through gestures.

Spidey and Rohan will go through the purpose of Andromeda, how Abi is driving this and how we incorporate machine learning into her physical communication.

Speaker Bios:
Adam Purdie Spidey, Mad Scientist / Engineering Manager.
Spidey regular meetup goer is best known for his ambitions to take over the world with an army of robotic spiders - furiously developing the spiders and other cuddly horrors by night, by day Spidey is an Engineering Manager in charge of customer applications and platform automation (SRE/Dev/Sec/Ops) and physical robotics. My team do awesome things and I give them all the direction, certainty and space to do it.

Rohan Nowell is a Machine Learning Engineer at Andromeda Robotics with over 10 years of experience in robotic and autonomous systems. His broad expertise spans control systems, mechanical engineering, physics modelling, and autonomy, including work on uncrewed defence vehicles and miniature robots from his PhD at Monash University. He is currently focused on the complex challenges of Human-Robot Interaction, specifically optimizing for 'delightfulness' in embodied AI systems.

Dushan Karovich-Wynne: “aigogo: Packaging AI Agents for Production”

Synopsis:
aigogo ( https://github.com/aupeachmo/aigogo )is an open-source project exploring what it actually takes to package, ship, and run AI agents in real production environments.

In this talk, I’ll walk through how to use aigogo to package and distribute a real agent that does useful work.

This talk isn’t about a finished framework, but an open exploration of what it takes to package and run AI agents in real production environments. Using aigogo as a concrete example, we’ll look at the practical tradeoffs that show up once AI systems move beyond experimentation and start behaving like deployable services.

The session is designed to be collaborative and discussion-driven, drawing on the experiences of engineers and data scientists in the room. The goal is to surface what matters most in practice and to help shape the roadmap for aigogo and similar open-source efforts going forward.

Speaker Bio:
Dushan is the founder of Subrosa AI (www.subrosa.ai), he is building at the intersection of AI, risk mitigation, and governance, with a focus on making AI systems secure, auditable, and trustworthy in production environments.

On top of his companies products which address data leakage and governance challenges as well as production-level AI agents, Dushan builds and maintains open-source projects that help people instrument, understand, sandbox and distribute AI agents.

Outside of work, he's a lifelong dog lover who has been known to pull over to talk to dogs.

Chaehan So: “Deep Listening AI: A Psychologist's Blueprint for Building a Virtual Friend”

Abstract:
What if AI could truly listen—not just respond with generic empathy, but recognize when you're avoiding a difficult topic, stuck in a thinking loop, or seeking validation you won't accept?
In this talk, psychological-researcher-turned-AI-startup-founder Chaehan So shares what building Virtual Friend revealed about creating AI for deep emotional conversations. You'll discover why common approaches like fine-tuning and RAG fail for psychological applications—and what actually works.
Key takeaways:

• Why "respond with empathy" prompts don't work (and what to do instead)
• How to engineer psychological awareness into AI conversations
• The hidden patterns that unfold across a conversation—revealing what someone can't say directly
  Whether you're building AI products, interested in mental health tech, or curious about the intersection of psychology and AI, you'll leave with practical insights applicable far beyond chatbots.

Speaker Bio:
Dr. Chaehan So bridges psychology and AI to create technology that genuinely understands people. As founder of Virtual Friend, he's building an AI companion specifically designed to support young adults navigating anxiety, loneliness, and life transitions—combining clinical listening techniques with cutting-edge language models.
With a Ph.D. in Social Psychology, Dr. So brings psychological research methodology to AI development. His interdisciplinary background spans social psychology (Humboldt University Berlin), engineering (TU Berlin), and management (ESCP France), with previous experience as Assistant Professor of Information & Interaction Design at Yonsei University, South Korea.
His mission: making AI that doesn't just talk to you, but truly listens.