Paolo Lombardi

In Nov 2012 I successfully led a team of 2 and me to raise €13 million from the local government in Trento (Italian Dolomite Alps region) to found a startup accelerator targeting international talent in digital, high-growth tech.

Within 6 months under my leadership, we:
- launched accelerator call and receive 600+ applications from 60+ countries (e.g. USA, UK, France, India, Russia, Belarus, Eastern Europe, etc)
- selected 63 participants for a 6-month residential program
- solved visa issues in Italy, and participated in co-creation of the Italian Startup Visa with the Ministry of Economic Dev.
- grew team to 12 and put it to speed
- lobbied to change a law on government-led startup funding
- brought Italian Prime Minister Enrico Letta to speak at the opening ceremony
- distributed €1 million funding to 24 startups
- set up and coordinated a top-notch mentor network including e.g. Bit.ly CEO and Joomla CTO
- organized a demo day with 200 investors

In 2 batches, we supported 45 startups in gaining product-market fit and traction, and raise total €12 million private funding. The accelerator had to close at end 2015 for exogenous reasons when we had deployed just 30% of funds.

From Aug 2015 to Feb 2016, I have been the product owner for the FundingBox renewed web platform, which was transitioning from a more traditional form-based UX to apply for being part of an SME consortium, to a more innovative Slack-like and user-centered communication platform named FundingBox Spaces.

The new positioning was the one-stop-shop web community for creating consortia to obtain EU funds. Our goal was to redesign the UX to empower users to create their own consortia, share information about funding opportunity in large groups, create private rooms as well as public discussion spaces, and enjoy a Slack-like experience.

I co-created the product roadmap, coordinated a team who did customer interviews, wrote user stories, prioritized features, and run continuous user tests as a product owner.

I interfaced mostly with the CTO based in Spain, the engineering team based in Poland, and other Directors and stakeholders located in New York, Copenhagen, Brussels, Warsaw, and Zaragoza.

In late 2011 together with a researcher in FBK, whom I had been supporting in my daily job in the previous 12 months, we spotted a market opportunity lying at the boundary of NLP and social media, where services in the USA were being created but only focused on English language support. Our technology could quickly be adapted to a range of 20+ different languages if its unsupervised learning algorithm was correctly supported by a human-created knowledge base.

We worked nights and weekends in stealth until we co-founded MachineLinking in Nov 2012, the premium service for short-text understanding and semantic disambiguation focused on Twitter and Facebook Feed conversations.

I acted as product manager for our NLP API which hit the market in mid-2013 and was adopted by a number of businesses worldwide sending language queries in English, French, Italian, German and Portuguese. I also contributed to the product strategy for our 1st vertical product, Pokedem, dedicated to Twitter streams of political candidates and their social managers.

Note: MachineLinking was closed in Dec 2014 (because of unsatisfactory operative margin) and the current Pokedem website is a relaunch/pivot of the original service and belongs to FBK.eu

In 2001, I boarded my first self-driving car and sat (freaking out) in the back seat as the car was self-speeding at 120 km/h on the motorway when my colleague took his hand off the wheel.

During my Ph.D. studies, I designed and coded in C/C++ and Java a number of algorithms for self-driving cars, including context-aware perception, Bayesian reasoning, road and pedestrian detection, time-series analysis, tracking filters, and so on.

In 2004-05 I brought that experience to a project of FCA (ex FIAT) and FBK Research Center. There I designed the computer vision system architecture, set up an onboard sensor and camera system (stereo camera set + independent camera), designed all computer vision algorithms, coded image processing to Bayesian systems.

I joined the project when it was in its last year of 3, and the previous engineer had quit without notice, leaving a poorly/non documented code which had not been tested nor even reviewed. In 12 months I merged with the existing project team of 6-7 people, recoded all that was (not) done the previous 2 years, and brought the vehicle to passing road tests successfully.

In May 2010, I became an advisor and early investor of a developer team in HFarm Accelerator in Venice, one of the Italian top startup accelerators and part of GAN (Global Accelerator Network), later (June 2011) to become the founding team of FunGo Studios, the first localization-based game studio in Italy.

Being a board game designer, I guided the product design of our first social mobile game: Gowar, a war game based on real-world maps and places.
Problems to solve included:
- UX for a blended real/virtual game
- battery life (GPS or wifi-based localization)
- real-world data exploitation to build the virtual world
- anti-cheating (e.g. jailbreaking)
- in-app purchase strategy
- maximization of in-game time
- AI player design

I created algorithmic simulations of the game dynamics in Matlab and designed the game mechanics related to pseudo-random battle resolution and the AI player.

Gowar outperformed expectations in the first 3 months after launch, with per-user in-app purchases peaking at 1000€.

After launch, I supported the creation of a community manager service and product fine-tuning and had a major role in designing our 2nd game Place Commander launched 2012. Both games have been discontinued.

Our partner was an international institution working with very secretive and highly sensitive surveillance systems, including cameras, volumetric sensors, and other rare kinds of sensors. The problem they were tackling was twofold: first, recreate temporal consistency of all sensor data flows; and second, making sense of these very diversified data streams with a clear UX/UI which could be used by human inspectors looking for anomalous events in the facilities under surveillance.

My role in the project was leading the product development, in a team composed of other 2 internal members and an external software company providing the UI+analytical software tool for one sensor-type. The goal was to create a higher level UI which would help spot inconsistencies between different data streams, to identify anomalies.

Initially, I ran user interviews and spent days on-site to observe how inspectors worked. Following my insights, we designed and coded an AI novelty detector based on HMM probabilistic event (generative) modeling of the underlying process. We integrated this AI in assisting mode with the existing analytics and the camera data in a new UI. The product was integrated into the client's standard procedure.

In late 2009 I started my first tech startup project, which ended up being a life lesson more than a startup.

An ex-colleague at FBK Research Center for AI and I, both scientists in computer vision and artificial intelligence, thought we spotted an opportunity in the home safety market of the aging society when seniors are more and more often living alone a long way from their grown-up children.

For 18 months, we co-developed an innovative multi-camera video surveillance system and filed two patents on highly efficient computable approximations of an NP-incomplete problem regarding tracking multiple people with occlusions in cluttered scenes. My role was acting CEO/Product Manager and my co-founder was acting CTO/Lead Developer.
I conducted:
- negotiation of software licensing from FBK Research Centre
- pre-seed fundraising (€150,000)
- team hiring (2 contractor engineers: hw and sw)
- market analysis and user interviews
- product roadmap
- MVP test
- business development
- business plan writing

Our efforts came to an end in summer 2011 when we couldn't prove market traction at a sustainable margin and agreed to close the project. Lessons learned:
1) customer interviews matter
2) product-market fit is tough

As Head of New Business Creation at Trento RISE, I was part of the core team who planned and designed one of the FIWARE Accelerators - FINODEX, specializing in tech startups working with Open Data.

As the Accelerator Manager at FundingBox, I supported the deployment of two more accelerators, namely IMPACT (on mobile tech and apps) and INCENSe (on energy tech and data).

My team and I successfully got the approval of the European Commission for our projects, did the setup of the accelerators start-to-finish, including positioning, selection process, mentor network, entrepreneurial training, funding deployment, acceleration activities, business development, communication, and PR, etc.

We worked hand-in-hand in total with 250+ startups (selected out of 1000s applicants) and deployed 20+ million Euros of funding to those startups. My role as a program manager was to coordinate my team in delivering the project, participate in startup selection and funding, coordinate mentors and jury members, and report to the European Commission.

I also coordinated consortia of stakeholders, including multi-national companies such as Enel or Docomo and problem-solved daily operations of the accelerators.

Driven the design and setup of an innovative and new Creative Industry Hub in Kuwait City in 2016-17, in a project for the Kuwait Government led by Lattanzio Group and the World Bank.

Got an incredible opportunity to work in this fascinating setting in the Middle East, with all the cultural charm and peculiarities of that world - just couldn't say no.

Soon after entering the project, I reviewed the strategy and the design that was originally conceived by Lattanzio Group to insert international best practices from players such as YCombinator, TechStars, and Startupbootcamp.

I then supported the launch of the call for startup and the selection process. Also, I coordinated the whole planning of program operations, selected the mentors, designed the content and its progression, set up progress milestones for startups and led the hiring of a project manager.

We received about 50 applications and selected 20 startups. Selected startups ranged in anything from artistic video making to hand-crafted jewelry - very far from digital and AI to which I belong. Low scalability was also a key characteristic of the batch.

I mentored personally the startups in Kuwait City on the business model and financial projections.