DeepJudge careers: remote jobs, culture and where they hire

DeepJudge has raised about $51M in two rounds. A $10.7M (€9.3M) seed round closed in June 2024, led by Coatue with angels including Gokul Rajaram and Michele Catasta, and an oversubscribed $41.2M Series A closed on November 4, 2025, led by Felicis (Viviana Faga and James Cheng joined) with continued support from Coatue, at a post-money valuation Forbes reported as $300M. Earlier backing came from Swiss angels including Felix Ehrat (former General Counsel of Novartis), Daniel Sauter (former chairman of Julius Bär), Thomas Hofmann (former Google director and ETH AI professor) and Thomas Dübendorfer (president of SICTIC), plus a CHF 150,000 Gebert Rüf InnoBooster grant in 2022 to 2023. At the Series A the company said revenue had grown by more than 500 percent over the prior year, and it is spending the money on product, its partner ecosystem and international expansion, with the US and UK the stated priorities.

DeepJudge builds the knowledge layer for legal teams: a retrieval-first platform that indexes a firm's document management system (iManage, NetDocuments, SharePoint, HighQ), email, intranet and client portals, then lets lawyers search it in natural language and run AI agents and workflows on top of it, without moving the data out of the firm's own systems. The product line has grown from Knowledge Search to AI Workflows (a library of pre-built workflows plus a low-code builder, launched March 2025), SuperSearch (cross-matter intelligence, launched March 24, 2026) and a ChatGPT Enterprise plugin, and it deploys into both cloud and on-premises environments holding hundreds of millions of documents.

Customers include Freshfields, Holland & Knight, Gunderson Dettmer (which retired its in-house ChatGD tool for it), Cozen O'Connor, ArentFox Schiff, CMS Switzerland, Schoenherr, Homburger and Lenz & Staehelin, plus Fortune 500 corporate legal departments. The company says its platform sees 80 to 90 percent active usage across its customer base, and attendees from 106 large law firms voted it the most recommended legal tech vendor in the SKILLS 2026 survey, its second year in a row at number one. It was named to the 2026 CB Insights AI 100, and in the last year has signed integration partnerships with Thomson Reuters (CoCounsel Legal), Harvey, Legora, Epiq, Microsoft Copilot and Perplexity, and published an open Agent Handoff Protocol that Harvey and Thomson Reuters were first to adopt.

Paulina Grnarova, CEO and co-founder. Researcher at Google in Zurich from 2017 to 2019 and a member of the Google Brain team, after an internship at ABB Corporate Research. PhD in computer science from ETH Zurich on deep generative models and GANs from a game-theoretic view (Ian Goodfellow sat on her committee), MSc in computer science from EPFL, bachelor's from the Faculty of Electrical Engineering and Information Technologies in Skopje. Forbes 30 Under 30 and a Digital Shaper of Switzerland three years running.

Yannic Kilcher, CTO and co-founder. Worked at Google AI Language on training, building and deploying ML models at scale. PhD from ETH Zurich on robustness of deep neural networks, NLP, reinforcement learning and generative models. Runs one of the largest YouTube channels on machine learning research (paper reviews and ML News) and contributed to the OpenAssistant open-source project. Grew up in northwestern Switzerland.

Kevin Roth, COO and co-founder. Physicist turned computer scientist who worked at Google Brain in Berlin and Microsoft Research in Cambridge, UK. PhD in machine learning from ETH Zurich, where he was awarded the ETH Medal in Physics. Runs operations at DeepJudge.

DeepJudge is a search company first and an AI company second, and that is the pitch to engineers: the founders came out of Google Brain and Google AI Language with a background in information retrieval, and the Series A post frames the whole product around it ("LLMs aren't so good at reading hundreds of millions of documents. But when you pair it with the search engine... the LLM can read the 20 and give you the answer," as CTO Yannic Kilcher put it to LawSites). The systems problems are real: dedicated production environments per customer, each a large-scale retrieval system running across multiple public and private clouds, with ingestion pipelines feeding hundreds of millions of documents.

The company is growing fast from a small base. LinkedIn data surfaced through Exa put headcount at about 90 in mid-2026, up 64 percent year over year, spread across six countries, and the careers page listed 22 open roles in September 2026 across engineering, product, sales, marketing, customer success and partnerships. Postings promise "ownership and ability to impact the company, product, and culture from day 1", and the Applied AI Engineer listing says plainly: "We care much more about your ability, judgment, and potential than about a specific number of years of experience."

Commercial hires sit close to the customer. Customer Success Managers own a portfolio of Am Law 100 or Magic Circle firms and run training and business reviews with senior partners; Enterprise Account Directors work alongside the VP of Sales with full ownership of a territory; Forward Deployed Engineers own a customer's deployment end to end, with no travel required.

The line every DeepJudge posting repeats is "we move fast, think rigorously, and care deeply about what we build", with newer listings changing it to "we move with urgency". Engineering postings add that the team values "engineers who care deeply about building high-quality systems while iterating quickly", and every listing closes with an invitation to apply even if you do not meet every qualification.

The founders describe the team in similar terms. CEO Paulina Grnarova, in a company Q&A: the people are "incredibly smart, talented, and motivated", and "we work hard, but we also have lots of fun together", with the team going to the sports centre together. Yannic Kilcher: "Our people are extremely competent and care a lot about their work. Together, we can tackle problems that nobody else even dares to touch." He also encourages firm leaders to "view technology as a method to enable your people to do their highest quality work, rather than as a cost item to be minimized", which is how the product is sold.

DeepJudge hires through an Ashby board, and there is a standing Open Application for full-time roles if nothing fits; working student and part-time roles are posted separately. Engineering listings in Zurich typically ask for a Master's in computer science or equivalent experience plus 3 or more years (4 or more for Platform Engineer), fluency with Python and a TypeScript/React front end, and comfort with Kubernetes; the Machine Learning Engineer role wants hands-on experience training and serving deep learning models in PyTorch, JAX or TensorFlow, with neural retrieval or agentic systems as a plus.

Commercial roles carry their own filters: US sales roles ask for 5 to 10 years of enterprise B2B sales, and the DACH Enterprise Account Director must be fully proficient in German and English and based in Zurich. Customer Success Managers need 5+ years in a client-facing SaaS or legal tech role, and a J.D. or time practising law is "highly preferred". Applicant data is kept for future openings unless you ask legal@deepjudge.ai to delete it.