Milos Colic — VP of Engineering @ Xoople | AI and Product Engineering Leadership | The first Earth data infrastructure company, built for AI
VP of Engineering @ Xoople | AI and Product Engineering Leadership | The first Earth data infrastructure company, built for AI
Milos Colic ranks #355 of 19,190 LinkedIn creators in Information Services, and is a standout voice in United Kingdom. They have 5.2K followers and published 58 posts in the last 30 days at a 0.6% average engagement rate.
- 5.2K followers
- 58 posts / 30d
- 0.6% avg engagement
- 446 follower growth / 30d
The roast
Milos claims to build data infrastructure for the real world, yet his content engagement rate is a rounding error away from not existing at all. He spends his days at Xoople trying to solve spatial gaps while his entire career profile remains stuck in the same place.
About Milos
I lead product engineering organizations, building trusted data and AI products with clarity, pace, and care. I am most useful when complex technical capability has to become something an organization can understand, trust, sell, and scale. That work usually sits between product, engineering, commercial reality, and judgment. The hard part is rarely the technology alone. It is deciding what matters, making the work legible, building the team that can carry it, and turning possibility into something people can depend on.At Xoople, I built the product engineering function from zero to one and now lead it as we build Earth intelligence infrastructure for AI.Before that, I led product engineering across CARTO’s core product surface, from developer platforms to end-user applications. At Databricks, I helped grow public sector and geospatial data opportunities in UK&I, worked closely with product and engineering teams, and created GeoBrix, a geospatial analytics framework adopted by hundreds of organizations. Earlier, at Barclays and Amadeus, I worked on regulated analytics, decision systems, large-scale data products, and patented NLP research.The thread through all of it is simple: technical depth only creates value when it changes decisions, systems, and outcomes. I care about teams that can think clearly under pressure, challenge theatre, build judgment, and leave behind systems stronger than any one person.I write about AI, product engineering, leadership, organizational transformation, and the practical work of turning technical possibility into real-world value.
Highlights
- Consistent Creator — 33 posts in 30d · top 5%
- Top 10% in Information Services — Ranked #4 of 76 creators
- Top 10% in United Kingdom — Ranked #108 of 1849 creators
- Top Engager — 0.56% rate · top 25%
Recent posts
Someone can now combine several specialised AI tools, pass an entire remote interview process and gain legitimate access to a company. It is fraud, obviously. But in a slightly disturbing way, the level of creativity behind it excites me. Not the fraud itself, but seeing how far we are already pushing the AI tools we have access to, particularly when several tools designed for specialised use cases are combined into something I am pretty sure none of their creators intended. Fraud may simply be the best illustration of the depth of that creativity. A synthetic identity is no longer just a fa
8 reactions · 0 comments · 0 reposts
Last night, I had the pleasure of spending the evening with the Databricks EMEA Digital Natives team and technology leaders from across the region. We had a great conversation spanning topics such as AI token economics, how AI is redesigning our organisations’ operating models under our feet, how hiring has come full circle, with pen and paper or a whiteboard once again becoming the best tools, AI optimism versus AI pessimism, and more. Above all, I left the room excited. Sharing and learning from people experiencing the same wave in a similar way, despite working across different industries
42 reactions · 2 comments · 0 reposts
I think we are now entering active AI/agent memory, rather than the passive memory systems we have been working with so far. What I mean here is that the memory layer is evolving into more than just data. It is not just about storing what happened, but learning how to change dynamic behaviour policies. This strongly resembles reinforcement learning, just not applied to weights but to memory, and perhaps even to other parts of the harness. Perhaps it could evolve the loops over time, making them implicitly task-specific and personalised to the consumer. Google Research’s ReasoningBank is wha
4 reactions · 0 comments · 0 reposts