Twenty-five years of building production systems, energy grids, battery platforms, analytics, retail media, fintech, AI. Some clients are public, some still stealth, a good number have exited. Here is the work.
Energy
E.ON, Virtual Power Plant
Energy · Germany, UK, Netherlands, Sweden · 2012–present
The problem. E.ON needed to dispatch heterogeneous energy assets, wind turbines, solar farms, CHPs, battery storage, across four countries in real time. TSO ancillary-service participation (aFRR in Germany) meant the system had to stay continuously available: an individual failure over 30 seconds was unacceptable. And the scope kept growing, trading across Intraday, Day Ahead, Wholesale and TSO ancillary markets at once.
What we built. The Virtual Power Plant, from proof-of-principle prototype to production, architecture, design, implementation, DevOps and the Agile transformation around it. A soft-real-time platform on Scala and Akka that models the virtual power plants, monitors and steers them, and adheres to the energy market's rules. Versatile enough that it manages car batteries as storage on the grid. We've run it with 24/7 L1/L2 support; the team scaled from 20+ down to a tight core of 7 as it matured.
The result. 99.99% uptime per annum, individual failures under 30 seconds. In 2025, close to 2GW of new installed capacity connected. At peak: 400 assets dispatched simultaneously, including a 400MW CHP. The same core team has been in this codebase since 2012, that continuity is the reason it holds up.
E.ON, Smart View II
Energy · enterprise EMS · 2014–2020
Scala · Java · Akka · Angular · iOS/Android
The problem. E.ON needed an Energy Management System for enterprise energy, consumption monitoring, reporting, forecasting, and a critical-information viewer, at the scale of thousands of buildings across Europe.
What we built. Version 2 of Smart View, front to back, across web and mobile.
The result. Selected as E.ON's EMS Lighthouse product, monitoring over 2,000 buildings across Europe.
E.ON, Optimum
Energy · machine learning · 2016–2018
Scala · Python · Go · Angular · machine learning
The problem. Turn data from intelligent meters, Building Management Systems and gateway boxes into energy intelligence, so E.ON's clients could cut consumption, cost and emissions, and decide faster.
What we built. Machine-learning models trained on smart-meter and BMS data, delivering energy intelligence into the hands of business decision-makers.
The result. Solutions that helped save energy and reduce emissions, cutting costs and carbon footprint for E.ON's business clients. An AI proof point that predates the current wave.
Sonnen → Shell, Battery Storage Platform
Energy · Germany, US · 6 years · through the Shell acquisition
Scala · Java · AWS · Kubernetes · GCP · Terraform
The problem. Sonnen needed two things at once: a residential battery-management system and an energy-contract fulfillment engine. The contracts were complex enough that off-the-shelf billing SaaS couldn't handle the calculation logic, and Sonnen was on a trajectory from AWS-serverless startup toward ISO-compliant enterprise, accelerated by Shell's acquisition.
What we built. The battery-management platform and a custom contract-calculation engine, every contract dimension calculated internally, then fed into the billing SaaS. We also helped execute the migration from startup infrastructure to ISO-compliant enterprise.
The result. A platform that survived acquisition and came out ISO-compliant and production-grade, built to hold up under Shell-level infrastructure review.
Over the years, Eloquentix has been a core driver in our ability to successfully launch and support numerous complex projects. They possess a rare "show up and build" mentality… they don't just execute on tickets, they advocate for the product, challenge our assumptions, care deeply about the long-term health of our tech. They operate with the ownership of an internal team. For any organization looking for a true engineering partner, Eloquentix is the gold standard. , Stephan Lindauer, Senior Director VPP Technology, Sonnen
SaaS & Analytics
Shoeboxed, 20x Growth
SaaS · Durham, NC · since 2008 · acquired 2018
Java · Scala · Play · MongoDB · PostgreSQL · AWS
The problem. A startup scaling from a few thousand users to over 800,000 businesses accumulates technical debt and architectural decisions that made sense at one scale and not the next. You need engineers who know where the bodies are buried, because they were there when the decisions were made.
What we built. Shoeboxed's receipt, business-card and document-management platform, continuously, since 2008. Not a project; an embedded relationship that outlasted every major architectural shift.
The result. 20x growth. 800,000+ business users. The most tenured engineer on the product was ours. Acquired by Earth Class Mail in 2018.
We consider our Eloquentix developers as much a part of the Shoeboxed team as their American counterparts. We have relied on them not only for their coding experience, but for their ability to help us solve a range of technology problems we experienced while scaling. , Taylor Mingos, CEO, Shoeboxed
Keen Decision Systems
Marketing analytics · SaaS · 11 years
Python · Angular · AWS · Bedrock
The problem. Keen set out to build a marketing-mix-modeling platform for major CPG brands, where the model is the product and credibility lives in the engineering. Building that and keeping it evolving for a decade-plus takes a team that doesn't churn.
What we built. Three Eloquentix engineers, backend, frontend, QA, embedded in Keen's product team since 2014. Roles and participation identical to any other Keen engineer, through the company's growth into a Series-B-funded analytics business.
The result. Eleven years, the same embedded team, continuously shipping, less a vendor relationship than a wing of Keen's own engineering org.
We have been working with Eloquentix since the start of our efforts to build a SaaS application. Our work spans over 10 years… they are seamlessly integrated into our organization, their roles and participation just as they are for any other engineer at Keen. They have consistently delivered productive, high quality work and advocated creatively for improvements to our product and development processes. I would highly recommend Eloquentix. , John Busbice, Co-Founder & Chief Decision Science Officer, Keen Decision Systems
The Looma Project, In-Store Retail Media
Retail media · in-store screens · ~7 years
Go · Python · Angular · Kotlin · GCP
The problem. Looma set out to build an in-store retail-media network, screens and an ad platform inside large grocery chains. A category-defining product needs foundational architecture that survives from seed-stage prototype to a system running across hundreds of retail locations.
What we built. The foundational architecture of the platform, built in partnership with Looma from early stage onward. The relationship went deep enough that an Eloquentix engineer is, to this day, the most tenured member of Looma's team, eight years in.
The result. Looma's network grew from 800 screens to 7,000+ across 1,100 retail locations, reaching 27M shoppers monthly in retailers including Kroger, BJ's Wholesale Club, Harris Teeter, Lowes Foods and Schnucks, and the company has raised $30M to date.
Eloquentix has been a phenomenal R&D partner. All of our foundational architecture was built in partnership with Eloquentix, and to this day one of their employees is the most tenured member of our team (8 years!). There's no chance we would be where we are today without Eloquentix. , Cole Johnson, CEO, The Looma Project
AI & Fintech
Augur, Autonomous Energy Forecasting
AI · energy forecasting · live on Predico since 2026
Python · probabilistic ML · conformal calibration · autonomous ops · AI agents
The problem. Europe's grid increasingly runs on weather, and every fifteen minutes the power market asks the same question: how much will wind and solar produce a few hours from now? Forecast errors settle in cash. The incumbent baseline, the grid operator's own continuously updated forecast, is genuinely strong, so beating it is the only honest test.
What we built. Augur, an unattended system that forecasts wind and solar production for Germany, Belgium and Romania and, since July 2026, competes on Predico's production instance, the collaborative forecasting platform operated by INESC TEC and used by Elia, Belgium's grid operator, submitting hourly probabilistic forecasts and self-scoring every one against the operator's own published forecast, including where it loses. A side project of ours, built in days almost entirely through AI agents, on twenty-five years of real-time energy-systems engineering. This is what Eloquentix looks like in the AI age: put AI on top of infrastructure we already ran for a decade.
The result. Live in production, scored against professional forecasters exactly as a trading desk would be. Short-range, Augur beats the operator's continuously updated forecast by 40–78% at fifteen minutes ahead, reaching parity around three hours out. Day-ahead, it beats every covered operator's wind forecast by 10–16% and runs +64% versus Transelectrica on Romania solar; prediction intervals hold about 80% empirical coverage at the 80% level. Hourly, zero missed gates. Where the operator wins, Augur says so and defers. See it run live, including where it loses, at augur.eloquentix.com.
Clockout, AI-Augmented Codebase Audit
Fintech · earned wage access · 2-day audit
Node.js/Express · React · PostgreSQL · Docker
The problem. Clockout runs an earned-wage-access platform white-labeled for community banks and credit unions, a fintech context where a code review can't be theater. They wanted a real read of the codebase, not a long report that flatters and obscures.
What we built. An Eloquent Lens engagement: cold start, no briefing. We fed the codebase to our AI-augmented audit pipeline and spent two days producing findings, a business-model reconstruction from code alone, an architecture assessment, scalability scenarios, security claims with specific references, and a git-history analysis that accurately described how the team works.
The result. A CTO who reviews code for a living found it useful precisely because it was uncomfortable where it should be, and committed to a follow-on build.
Eloquentix came in cold. No briefing, no architecture walkthrough, no README tour. They fed the codebase to an AI-augmented audit pipeline and spent two days producing findings… a git-history analysis that accurately described how my team works. That last one was the tell. If you can read a team's git process and get it right, you actually read the code… I'll be bringing them back. We have more to build. , Vladimir Dumitrean, CTO, Clockout
JustWin
GovTech · AI proposals · scaled to 300+ customers
Python · Angular
The problem. JustWin builds an AI platform for government contract bids and proposals, a market that demands two things that usually trade off: ship fast enough to win a fast-growing customer base, and meet the security bar that sensitive government-adjacent data requires.
What we built. A product-engineering partnership that turned priorities into shipped features at startup speed while holding security-and-reliability discipline appropriate to the data, through a scale from a few dozen customers to 300+.
The result. 300+ customers and counting; the founder cites execution speed as a direct accelerator of growth.
Eloquentix has been an invaluable partner as we've scaled our business from a few dozen customers to 300+. What stands out most is their ability to move quickly and ship… that execution speed has been a major accelerator for our growth. At the same time, they're seasoned professionals who care deeply about security and reliability. That combination is rare, and exactly what we need. , Ted Baxa, CEO & Co-Founder, JustWin
A Founder's First Software Product
AI-first product · founder partnership · stealth
The problem. A founder from outside software, a real-estate investor and dealmaker, arrived with a personal, ambitious vision for an AI-first product, and less experience in the domain than in his own. He needed a partner who would tell him the truth, including the uncomfortable financial parts, before he spent.
What we built. We led with the risk: in the first meeting we flagged the financial exposure up front. From there, a collaborative build that stayed on the cutting edge as the technology moved, and respected the deeply personal vision behind the product.
The result. A finished product that exceeded the founder's original expectations, and a relationship that outlasted the engagement.
From our very first meeting, I was impressed. The CEO immediately alerted me to the financial risks involved right out of the gate. This level of honesty and transparency earned my trust instantly… we created a final product that far exceeded our original expectations… they are now more than just service providers, they are friends. , Founder, stealth AI product (real-estate background)
Platforms & Long Engagements
NextLot, Real-Time Auctions
Online auctions · live video + bidding · since 2012
Ruby · Flash · Java · AWS
The problem. A white-label auction platform inherited broken and incomplete, buckling under load. Beyond stabilizing it, the hard ask: live video synchronized with web bidding in real time. Video lag accumulates over a live stream; keeping it in sync with auction bids breaks most teams, and in 2012 there was no playbook.
What we built. We fixed the platform, then went past the brief and built the synchronized live-video-plus-bidding system, solving the drift problem that makes real-time auction video hard.
The result. A working real-time auction webcast, delivered under pressure.
We inherited a broken and incomplete platform… Eloquentix came in, fixed it, and then pushed further. They built live video synchronized with web bidding. In 2012, that was genuinely hard to pull off… the kind of problem that breaks most teams. They solved it. Most vendors talk about what they can do. These guys just showed up and built it. , Matt Warren, then at NextLot
Scuttlebutt Research
Market intelligence · institutional investors
Python · AWS
The problem. Scuttlebutt runs a market-intelligence platform for institutional investors, a product where performance and reliability are the feature. They needed to stabilize performance and keep shipping without slowing down.
What we built. Embedded engineers who stabilized the platform's performance and drove ongoing feature development, integrated directly into the existing dev flow.
The result. On-time delivery, sustained, the kind of integration where the line between vendor and team disappears.
Eloquentix team is world-class in both code quality and professionalism. They've repeatedly delivered on-time results and have integrated seamlessly into our dev flows. Truly an extension of the team. , Will Thompson, CEO, Scuttlebutt Research
Clear Channel, Salesforce, 11 Years
Media · Salesforce / enterprise CRM · since 2012
Salesforce · Angular · Bootstrap
The problem. A global outdoor-advertising company running critical business functions on Salesforce needed customized solutions and steady administration of internal and customer-facing environments, work that compounds in value when the same people hold it for years.
What we built. Customized Salesforce solutions supporting critical business functions, plus system administration of internal and customer-facing Salesforce environments, the same small embedded team since 2012.
The result. Over a decade of continuous delivery with zero account turnover, deep domain continuity that only a stable embedded team produces.
ndustrial.io, Nsight
Industrial IoT · energy management · since 2013
Ruby · React · Python · AWS
The problem. Industrial energy monitoring and optimization for large retailers and plants, turning real-time site data into efficiency and cost/emissions reduction.
What we built. Core platform engineering on Nsight, ndustrial's energy-management system, full-stack and DevOps, embedded over a multi-year engagement dating to 2013.
The result. A production energy-management platform serving industrial and large-retail sites; ndustrial has since raised a $6M Series A (2021) and an $18.5M Series B (2024, co-led by ABB and GS Energy).
Liine
Dental / healthcare SaaS · 6 years
Python · Angular · ML
The problem. Liine helps dental and healthcare practices turn inbound calls into booked patients, lead capture and conversion analytics where accuracy and uptime drive the customer's revenue.
What we built. An embedded engineering team on Liine's core product, sustained over six years.
The result. A six-year partnership, the kind of retention that only comes from a team treated as part of the product org.
PayRange (Vagabond)
Vending · mobile payments · acquired 2023
Node.js · Angular · PostgreSQL · AWS → GCP
The problem. Vending operators needed real-time business intelligence and frictionless mobile payments, inventory, route planning, and a payments platform that could fail over gracefully.
What we built. The payments platform from the ground up, multiple payment integrations running standalone or in parallel for failover, real-time remote monitoring of every transaction, and the AWS-to-Google-Cloud migration that carried it.
The result. A platform that became the operating system for many SMB vending operators, and was acquired by PayRange in 2023.
Energy is the deepest bench Eloquentix has. Beyond E.ON's Virtual Power Plant and Sonnen's residential battery platform, the same domain expertise runs through Autogrid (flexibility management, since acquired by Schneider Electric), BitWatt (retail-energy pricing and procurement), and WindSim Power (wind-farm simulation). Real production systems steering actual power assets across European markets, not proofs of concept.
Where they ended up.
It started with Extensibility. Dot-com days, a small team, a Series A, and barely a year later TIBCO bought it. That's where the startup bug bit, and it never quite let go.
Twenty-five years on, that's still the work: be the engineering team behind someone else's company, do the part that's hard, and hope the next milestone is theirs to celebrate. A good number of them got there.
