At a Glance: Maneva builds Video-to-Action AI for the factory floor, turning ordinary cameras into real-time quality control and line supervision. Moving that from research into production needed six specialists across three roles at once: an AI software engineer, four data annotators, and an application delivery engineer. Second Talent filled all three briefs from a single pipeline, the AI engineering role in 27 days. Three months after the last offer was accepted, Maneva announced a $27 million Series A.
Company Background
Maneva is an industrial AI company founded in 2021, with offices in Palo Alto, California and North York, Ontario. Its Video-to-Action platform takes the feed from off-the-shelf cameras already installed on a production line and converts it into decisions a plant can act on in real time.
The product line runs on two systems. VITA handles defect detection and quality assurance, catching faults as parts move rather than at an end-of-line inspection station. ALIS, the AI Line Supervisor, oversees production flow, tracks efficiency, and monitors safety compliance across the line.
The founding team came from the research side of exactly this problem. Co-founder and CEO Rae Jeong was a research engineer at Google DeepMind working on AI and robotics. Co-founder and CTO Kelvin Chan was an R&D engineer at Magna, building perception technology in computer vision, RF, and deep learning for automotive and manufacturing. The company raised a $10 million seed round in 2024 to scale deployments, and now sells into automotive, electronics, food and beverage, steel, and pharmaceutical manufacturers.
Challenges Faced
1. Three unrelated skill profiles, needed at the same time
AI software engineering, data annotation, and application delivery are three different labour markets with three different vetting standards. A seed-stage team hiring all three at once through separate channels ends up running three searches, each with its own timeline and its own bar for what “good” looks like.
2. Annotation capacity is the hidden constraint in computer vision
A defect detection model is only as good as the labelled video behind it, and that work does not stop once the model ships. Every new customer line, part, and failure mode adds more to label. Annotation is usually treated as an afterthought or pushed to a faceless vendor, which is how label quality quietly becomes the ceiling on model accuracy.
3. Competing for AI engineers against Palo Alto and Toronto benchmarks
Maneva’s two offices sit in two of the most expensive engineering markets in North America, and it was hiring against companies with far deeper pockets for the same computer vision skill set. Adding six specialists locally would have consumed a disproportionate share of a seed round meant to fund deployments.
Solutions Delivered by Second Talent
Second Talent ran all three briefs through one pipeline of pre-vetted candidates, applying a single standard across the AI engineering, annotation, and application delivery roles rather than treating each as a separate search.
The data annotation brief was the largest, at four hires, and it was the first opened and the last to close, which says something about the role that its title does not. Annotation for industrial computer vision is not general-purpose labelling. It needs annotators who can tell a real surface defect from a reflection or a shadow on a moving line, and getting that judgement wrong at the labelling stage caps how accurate the model can ever be.
Speed mattered most on the engineering side, where the roadmap was waiting. The AI software engineering brief went from open to accepted offer in 27 days, and the application delivery role closed inside six weeks. Neither search ran long enough to become a second full-time job for a founding team already shipping to customers.
Companies running the same play can hire AI and machine learning engineers through the same pipeline, or use AI staffing to cover several roles at once.
Results Achieved
Maneva filled all three briefs with six hires, covering the full path from training data to production software without expanding headcount in Palo Alto or Toronto. Each of the three roles reached an accepted offer, and the team was in place inside a single quarter.
That mattered because the roadmap was the constraint. VITA and ALIS were already in front of customers, and every new plant, part, and failure mode meant more video to label, more model tuning, and more product surface to ship. Adding capacity at all three of those points at once is what let the product keep moving instead of queueing behind a hiring process.
What the platform delivers in the field
Maneva reports the following results from customer deployments of the two products these roles support:
On that wood line, stoppages went from one every 1.5 minutes to one every 25 minutes. ALIS deployments report roughly a 50% improvement in safety performance and up to a 10% increase in output through higher worker productivity.
A $27 million Series A three months later
In June 2026, three months after the last offer was accepted, Maneva announced a $27 million Series A led by U.S. Venture Partners, with returning investors Bling Capital and Freestyle Capital. It took the company total funding to $38.4 million.
The round is earmarked for deepening VITA and ALIS, expanding into new markets and industries, and launching factory orchestration and AI knowledge agents. That is model work, training data, and application delivery, which is the same three-way split the team hired for a quarter earlier.
Team Composition
Key Outcomes
- AI engineering role filled in 27 days: Open brief to accepted offer, with the application delivery role closing inside six weeks.
- Three briefs, six accepted offers: An AI software engineer, four data annotators, and a mid-level software engineer for application delivery.
- The full AI delivery stack covered: Model engineering, training data, and the software that puts both in front of a plant operator.
- Annotation hired as an engineering role: Labelling capacity treated as part of the model team rather than outsourced away from it.
- One vetting standard across three labour markets: A single pipeline instead of three parallel searches with three different bars.
- Specialist capacity without a Palo Alto cost base: Seed funding kept pointed at deployments rather than at local headcount.
- A $27 million Series A three months later: Led by U.S. Venture Partners in June 2026, taking Maneva total funding to $38.4 million.
Build a team like this on Second Talent
Browse pre-vetted senior AI-native engineers, shortlist, and hire on one platform. Start free, no upfront fees.
Try for Free
