<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Sophie Meralli on Medium]]></title>
        <description><![CDATA[Stories by Sophie Meralli on Medium]]></description>
        <link>https://medium.com/@sophie.meralli?source=rss-e19e637ccf0c------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*p1pWdgNXOCX7DTlrxNELRw.jpeg</url>
            <title>Stories by Sophie Meralli on Medium</title>
            <link>https://medium.com/@sophie.meralli?source=rss-e19e637ccf0c------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 03:16:54 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@sophie.meralli/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Routable AI and the Ridesharing 2.0 Era]]></title>
            <link>https://medium.com/innospark-ventures/routable-ai-and-the-ridesharing-2-0-era-73dab9102028?source=rss-e19e637ccf0c------2</link>
            <guid isPermaLink="false">https://medium.com/p/73dab9102028</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[ridesharing]]></category>
            <category><![CDATA[lyft]]></category>
            <category><![CDATA[mobility]]></category>
            <category><![CDATA[uber]]></category>
            <dc:creator><![CDATA[Sophie Meralli]]></dc:creator>
            <pubDate>Mon, 11 Nov 2019 16:56:11 GMT</pubDate>
            <atom:updated>2019-11-11T16:56:11.673Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*xFEY_e8DCQXlr21UJDsbfg.gif" /><figcaption>Routable AI’s revolutionary pooling algorithm in action</figcaption></figure><p><strong><em>Routable AI is enabling the next wave of the ridesharing revolution. With its AI-driven operating system for high-capacity fleets, it will provide efficient and convenient low-cost transit for daily commuters across the world.</em></strong></p><h4><strong><em>The rise and limitations of ridesharing 1.0</em></strong></h4><p>The global ridesharing market is gaining popularity and is expected to grow at <a href="https://www.marketsandmarkets.com/Market-Reports/mobility-on-demand-market-198699113.html">a CAGR of 19.87% from 2018 to 2025, to reach a market size of $218.0 billion by 2025</a>, driven by growing population and rising urbanization in emerging economies such as China and India.</p><p>Yet, ridesharing in its current form, or <em>Ridesharing 1.0</em>, <strong>is still far from ubiquitous. </strong><a href="https://www.mckinsey.com/industries/automotive-and-assembly/our-insights/cracks-in-the-ridesharing-market-and-how-to-fill-them">McKinsey research</a> indicates that it represented only ~1% of total vehicle miles traveled (VMT) in the US in 2016 and penetrated only about a third of passenger use cases.<strong> Several limitations, including technology and affordability, prevent ridesharing from scaling to more use cases. </strong>For use cases such as working commutes (13% of underserved US VMT) or rural areas (25% of underserved US VMT), their research shows a 6–8 seating capacity may be more comfortable and affordable, but due to its technology limitations, Ridesharing 1.0 cannot service high-capacity fleets without generating significant wait times and delays. Typically, current ridesharing algorithms are “greedy”, which means the nearest vehicle gets scheduled rather than optimizing across the network.</p><p><strong>As a result, Ridesharing</strong> <strong>1.0</strong> <strong>remains more expensive than owning a car for the overwhelming majority </strong>and<strong> </strong>has not lived up to its promise to reduce vehicles on the road and make traffic better. Instead, it has had a <strong>negative impact on public transportation</strong>. <a href="https://www.citylab.com/transportation/2019/01/uber-lyft-ride-hailing-impact-public-transit-ridership/581062/">For every year after ridesharing companies enter an urban market, studies show that rail ridership falls by 1.3%, and bus ridership by 1.7%. Thirty-one of 35 major metropolitan areas in the US lost public-transit passengers in 2017</a>. Higher disposable incomes and higher demand for convenience have also exacerbated the challenges of public transit including first &amp; last mile connectivity issues and irregularity, leading to<strong> public fleet under-utilization</strong>.</p><p>In addition, <a href="https://www.theverge.com/2019/8/6/20756945/uber-lyft-tnc-vmt-traffic-congestion-study-fehr-peers">recent studies </a>indicate that <strong>far from alleviating traffic congestion</strong>, ridesharing 1.0 led to <strong>increased travel times in cities</strong> due to vehicles cluttering the streets and a practice known as <strong>deadheading</strong> — where ridesharing cars spend about <strong>a third of their miles traveled with no passenger.</strong> Another independent study commissioned by the San Francisco County Transportation Authority looked at 2017 traffic patterns in the county and showed that <a href="https://www.theverge.com/2019/5/8/18535627/uber-lyft-sf-traffic-congestion-increase-study">between 2010 and 2016 traffic congestion in San Francisco increased by about 60% — and Uber and Lyft are responsible<strong> for more than half of that increase</strong>.</a></p><p>In order for ridesharing to deliver on its potential public benefits — including reduced energy consumption and greenhouse gas emissions, easing of road congestion, as well as affordability and accessibility — it will need to evolve and employ advanced technology like AI to <strong>enable more efficient fleet utilization </strong>with <strong>higher occupancy vehicles</strong> and <strong>favorable economics</strong>.</p><h4><strong><em>Meet Routable AI — a dynamic operating system enabling Ridesharing 2.0</em></strong></h4><p>Routable AI is an advanced routing engine that enables ridesharing to scale to high-capacity fleet of 12+ riders per vehicle, with efficiency and convenience — a constraint optimization problem exponentially more complex than the pooling of 4-passenger vehicles.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*yMH6NMh8T0m31nhDOf6hkQ.jpeg" /><figcaption>Menno Van der Zee (co-founder &amp; CEO)</figcaption></figure><p>When <a href="https://www.linkedin.com/in/mennovanderzee">Menno Van der Zee</a> and <a href="https://www.linkedin.com/in/wallarelvo">Alex Wallar</a> realized they were working on the same optimization problem for mobility-on-demand systems at two different institutions, Delft University of Technology and MIT CSAIL, they initially intended to prove each was better than the other, but in so doing realized they were better off teaming up. That is what led them to co-found Routable AI.</p><p>The company uses a novel AI approach that combines cutting-edge constraint optimization models with performance engineering to make it possible to quickly iterate through millions of possible trip combinations and dynamically route fleets of over 5,000 high-capacity vehicles<strong>, while guaranteeing all passengers a quick and efficient ride</strong>. Their initial <a href="https://www.pnas.org/content/114/3/462/tab-figures-data">MIT research published in PNAS</a> demonstrated that they could service all of Manhattan’s taxi demands with only ⅕ of the vehicles, a passenger wait time of fewer than 3 minutes and without putting the passengers more than 6 minutes out of their way due to carpooling.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*M37ViiyCO3LUsFu92MDzgg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5Ws13Bbu_zZRpgU0QS7Vcg.png" /><figcaption>Routable AI Fleet Management Dashboard</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vMCspwciVm-6hNyeSPyuTw.jpeg" /><figcaption>Alex Wallar (co-founder &amp; CTO)</figcaption></figure><p>Routable AI is now partnering with municipal transport operators to introduce on-demand services, and with private operators and ridesharing companies to better manage their vehicle networks. Their plug-and-play fleet management software integrates with <strong>existing fleets</strong> across the world and provides efficient routing, passenger pooling and vehicle rebalancing.</p><p>This model will alleviate many of the problems that ridesharing 1.0 has created. It combines the <strong>low-cost structure </strong>and <strong>efficiency</strong> of public mass transit with the <strong>convenience</strong> of private end-to-end individual transit systems, and offers the services at a similar price point to public transit, thereby increasing <strong>access to lower income commuters</strong>.</p><h4><strong><em>On-demand mass transit and the future of commuting</em></strong></h4><p>With the advent of ridesharing, there has been a massive shift underway in how people perceive ground transportation, a transformation that could impact the growth paradigm of the automotive industry and public transit.</p><p>We envision a world where public and private transit will become flexible and optimized, powered by AI-driven solutions. Buses may follow a fixed route during peak hours, and an on-demand model in non-peak hours. Commuters in underserved areas where public infrastructure planning is lagging will be able to commute with ease through inexpensive on-demand services.</p><p>We are excited about the future of commuting and the public benefit of this new model, <strong>improving efficiency of the system, lowering the cost of transportation, reducing congestion and emissions, and delivering convenience</strong>.</p><p>We are thrilled to welcome Routable AI to our portfolio and to help them steer our planet towards a more sustainable state.</p><p>Check out their <a href="https://routable.ai/">website</a> for latest info!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/646/1*67hj2sg5GYj0KCnKs7NcOw.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=73dab9102028" width="1" height="1" alt=""><hr><p><a href="https://medium.com/innospark-ventures/routable-ai-and-the-ridesharing-2-0-era-73dab9102028">Routable AI and the Ridesharing 2.0 Era</a> was originally published in <a href="https://medium.com/innospark-ventures">Innospark</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From aerial data to CAD drawings in minutes: the story of Airworks]]></title>
            <link>https://medium.com/innospark-ventures/from-aerial-data-to-cad-drawings-in-minutes-the-story-of-airworks-82eea3ad372d?source=rss-e19e637ccf0c------2</link>
            <guid isPermaLink="false">https://medium.com/p/82eea3ad372d</guid>
            <category><![CDATA[construction]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[civil-engineering]]></category>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Sophie Meralli]]></dc:creator>
            <pubDate>Wed, 23 Oct 2019 03:08:23 GMT</pubDate>
            <atom:updated>2019-10-23T03:09:16.044Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*C8CcDmrIiEIwINgCUbNIJA.png" /></figure><p><strong>AirWorks uses computer vision and deep learning to analyze aerial data in order to deliver CAD linework and topography to civil engineers and land surveyors, at a fraction of the time and cost of traditional surveying methods.</strong></p><p><strong><em>Two co-founders with a bold vision to automate the processing of aerial data</em></strong></p><p>When we met with David Morczinek and Adam Kersnowski, the founders of AirWorks, we were impressed by their bold vision and passion to drive substantial time and cost savings into the lesser known $8 billion US Land Surveying industry, thereby disrupting the existing industry structure. Some of us may have encountered at some point in our lives a surveyor on the streets taking GPS and optical measurements with a tripod — a labor-intensive and time-consuming practice. This traditional practice has not fundamentally changed in hundreds of years and is now ripe for AI-based disruption. Our interviews with civil engineers and land developers revealed that surveyors are overbooked and can take several months to complete a survey project. In addition, Surveyors are getting older and new young talent is not being recruited into the profession. The average age of surveyors is between <strong>58 and 60 years</strong>. While there are many players increasingly turning to unmanned aerial vehicles (UAVs) for autonomous data collection and image stitching, and various startups striving to serve them, no one other than Airworks is developing an <strong>aerial intelligence</strong> solution that converts aerial data into multi-layered CAD drawings.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/729/1*zmo1SxtZRChhiseYemtu0w.png" /></figure><p>David and Adam decided to combine their complementary skill sets to automate CAD drafting using AI, thereby speeding up the current surveying process by weeks. In order to succeed, they needed sophisticated geospatial, computer vision and civil engineering expertise. David has a background in aerospace engineering and managed aircraft sales and manufacturing contracts during his time at Airbus. During his MBA at MIT Sloan, he immersed himself into the UAV startup ecosystem, where he met Adam. Adam had run multiple businesses (including construction and UAV piloting) for the past 10 years. Together, they bring the perfect mix of domain expertise, engineering, and business skills required to achieve their vision.</p><p>Over the past 8 months, AirWorks has scaled from MVP to launching their SaaS product offering, at a remarkable pace. <strong>AirWorks’ proprietary platform now autonomously converts 2D and 3D aerial data into CAD models within two hours. </strong>Their goal is to drive the processing time down to a few minutes. The generated CAD files are available in common formats (.dxf, .dwg. etc.) and are ready to use in standard design software packages such as AutoCAD.</p><p><strong><em>Toward a scalable approach to AI: the importance of human-machine collaboration to raise contextual awareness in autonomous processes</em></strong></p><p>David and Adam realized early on that executing on their vision would come with challenges: for the AI-powered solution to be commercially viable and adopted by customers, it had to provide industry-grade accurate measurements and fit into current workflows.</p><p>Needless to say, deep learning has shown impressive results in the field of computer vision and is subject to fast pace continuous improvements. However, these models have not been trained for specific applications such as civil engineering, and to date, can provide correlations based on large amount of data but do not have the necessary contextual understanding to rely entirely upon them. David and Adam concluded early on that human-machine collaboration was needed to provide satisfying level of accuracy and ready-to-use CAD drawings. They moved away from the strict black box approach, towards understanding the civil engineering nature of the work and incorporating several abstractions and structures into their machine learning stack.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vrbxoYzBhLvc6wJ4iwFqUw.png" /></figure><p>To achieve this, the founders set out to build a strong interdisciplinary engineering team with extensive knowledge of land surveying, construction, civil and geospatial engineering, computer vision and deep learning, and software architecture. The founders established a culture in which every employee is aligned and incentivized to achieve the same company milestones. This created unity and commitment within the team from the beginning and fostered organic cross-functional collaboration.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*246Ud8IS7DGnx_wYqW9Ovw.jpeg" /></figure><p>We love to stop by their office and see<strong> the camaraderie and collaboration as civil engineers and AI engineers work side by side to improve the accuracy of their algorithms and combine computational thinking and domain expertise into aerial data analysis.</strong> “It is vital to communicate with and understand the needs of our customers. That is why we work closely with our team of civil engineers to collect the best possible data so that we can deliver the best possible product to our customers” — says Brian Bongiovanni, Computer Vision Engineer at AirWorks.</p><p><strong><em>What next?</em></strong></p><p>With the AirWorks SaaS platform, ordering CAD drawings with linework of topography and existing conditions for construction and engineering projects, is just a few clicks away. Key benefits for civil engineers and land developers include higher productivity, faster project cycles, increased profitability and better cost planning.</p><p>The waitlist for access to the AirWorks autonomous drafting software platform is now officially open! The first 50 customers to join the waitlist get 100 acres of data processed for free.</p><p>Join the <a href="http://airworks.io/airworks-signup">waitlist</a> or check out their <a href="http://airworks.io/">website</a> for the latest info!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sOoupZ7io385SeIV0Xc1mw.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=82eea3ad372d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/innospark-ventures/from-aerial-data-to-cad-drawings-in-minutes-the-story-of-airworks-82eea3ad372d">From aerial data to CAD drawings in minutes: the story of Airworks</a> was originally published in <a href="https://medium.com/innospark-ventures">Innospark</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From MIT Sloan to VC: My Journey as an International Student]]></title>
            <link>https://medium.com/@sophie.meralli/from-mit-sloan-to-vc-my-journey-as-an-international-student-7e56d339ae?source=rss-e19e637ccf0c------2</link>
            <guid isPermaLink="false">https://medium.com/p/7e56d339ae</guid>
            <category><![CDATA[immigration]]></category>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[boston]]></category>
            <category><![CDATA[international]]></category>
            <dc:creator><![CDATA[Sophie Meralli]]></dc:creator>
            <pubDate>Mon, 09 Sep 2019 18:28:55 GMT</pubDate>
            <atom:updated>2019-09-09T18:48:47.132Z</atom:updated>
            <content:encoded><![CDATA[<p><a href="https://www.linkedin.com/in/sophie-meralli/"><strong><em>by Sophie Meralli</em></strong></a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bjoWmIHZG_VbmlQ9J9NY2Q.jpeg" /></figure><p>After completing my MBA at MIT Sloan last year, I was excited to join <a href="http://www.innosparkventures.com/">Innospark Ventures</a> — a newly founded early-stage VC based in Boston. Being passionate about entrepreneurship, I thought it’d be great to gain experience in a leading technology startup eco-system like Boston. I started working under a 1-year OPT program, without a clear visa strategy for the following year. Unlike, big corporations that have immigration resources for talent retention, my team was only 2 people at that time, and I had to look into visa options such as O1, green card, cap H1B, etc. I concluded the most relevant solution to my case would be to apply for the H1B lottery with a Master cap. The chances to get selected were quite limited, and even with that, there were a lot of cases of denial due to the recent rise in requests for evidence.</p><p>This issue was faced by my international classmates, also pursuing careers in entrepreneurship and starting their own companies. As I began sourcing deals, I became excited about <a href="https://airworks.io/">AirWorks</a>, an aerial data imaging company started by my former classmate, David Morczinek. It turned out David and one of his first employees had the same immigration uncertainty. Employers can sponsor H1B but this implies that the startup needs to have some operations in the US before applying and that an entrepreneur cannot be his own employer for the purpose of this visa. Starting a business involves dealing with many risks already on the business side, and this visa situation worsens the startup risk profile, jeopardizing ability to raise funds and attract talent. For VCs like Innospark, this meant there was a high chance David, CEO and co-founder of the company, would need to leave the country the following year. For these reasons, international entrepreneurs have difficulties starting their business in the US. In some countries, there are visas designed specifically for startups and entrepreneurs. For instance, France, my home country, has put in place a “<a href="https://visa.lafrenchtech.com/">French Tech Talent Visa</a>”, which costs ~$400 and allows eligible founders and VC investors to stay in France for 4 years. In the US, currently there is no such solution as an official entrepreneur visa.</p><p>Fortunately, for David and me, we heard about the <a href="http://vdc.umb.edu/entrepreneur-visa-sponsorship/"><strong>Global Entrepreneur in Residence (GEIR) program</strong></a> at the <a href="http://vdc.umb.edu/">Venture Development Center</a>, UMass Boston. The GEIR is a program that provides cap-exempt H1B immigration solutions to global talent pursuing a career in entrepreneurship (startups/VCs). Members of the program work part-time at the VDC as mentors for startups of the VDC and broader UMass Boston community. UMass Boston is one of the first universities taking part in this initiative led by the Commonwealth of Massachusetts, and now adopted in six states and 11 universities.</p><p>The GEIR program enabled me to remain in the US after the end of my OPT. As part of the program, I mentor other VDC startups on crafting a compelling investor pitch deck, approaching investors, and elaborating their go-to-market strategy, AI and product roadmap and hiring plans. In my opinion, the GEIR program is much more than an immigration solution — <strong>it is a unique community of talented entrepreneurs, professors and students</strong>. Early-stage entrepreneurs can benefit from hands-on mentorship from fellow early-stage founders and VC investors that are part of the program. As a Global EIR at the VDC, I am excited to engage with the vibrant ecosystem UMass Boston has created and to advise early-stage entrepreneurs with fundraising and business development.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7e56d339ae" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>