Including Funding rounds, Bull / Bear thesis, Stock + earnings, Roster changes, Patents, News, and Open roles.
Already subscribed? Sign in →
Vladislav (Vlad) Voroninski is a mathematician-turned-founder. He graduated summa cum laude in applied mathematics from UCLA and earned his PhD in mathematics at UC Berkeley in 2013 under Emmanuel Candès and John Strain; his thesis introduced the PhaseLift algorithm and the first mathematical recovery guarantees for phase retrieval, effectively launching that field within applied math. He served on the MIT mathematics faculty from 2013 to 2016 and was chief scientist at Sift Security, a machine-learning cybersecurity startup acquired by Netskope in 2018. In 2016 he co-founded Helm.ai with Tudor Achim to apply insights from compressive sensing and deep learning theory to autonomous driving, developing an unsupervised-learning approach the company calls Deep Teaching that dramatically cuts the cost of training perception systems. As CEO he has steered Helm.ai toward supplying AI software to automotive OEMs and Tier 1s rather than operating vehicles, raising a $55M Series C in August 2023 with strategic backing from Honda, Goodyear Ventures, and Sungwoo Hitech.
Tudor Achim's path ran through competitive piano and computational biology before machine learning: he worked on ML at Quora and was a PhD candidate in computer science at Stanford before leaving academia to build startups. In 2016 he co-founded Helm.ai with mathematician Vladislav Voroninski, serving as CTO and helping develop the company's Deep Teaching unsupervised-learning approach to autonomous driving perception. Achim left Helm.ai and in 2023 co-founded Harmonic with Robinhood CEO Vlad Tenev, where he is CEO. Harmonic builds mathematically grounded AI for verifiable reasoning; in July 2024 its system achieved gold-medal-level performance on International Math Olympiad problems, matching results from OpenAI and Google DeepMind, and the company has since raised a $120M round.
No articles ingested yet for Helm.ai. Once the hourly news pipeline is live, every article the classifier tags as mentioning this company appears here with its one-line AI summary and sentiment.
A camera-based perception stack for ADAS and autonomous driving trained using Helm.ai's Deep Teaching unsupervised-learning method, which extracts patterns from large unlabeled datasets rather than relying on costly manual annotation. Helm.ai Vision provides production-grade object detection, semantic understanding, and scene perception that automakers can deploy across Level 2 to Level 4 systems. The unsupervised approach is the company's core differentiator, aimed at delivering high-accuracy perception while reducing the data-labeling cost that dominates conventional autonomy development.
A vision-based, real-time path-prediction and driving-policy system that turns perception output into safe vehicle behavior for hands-free and automated driving. Designed to run on automotive-grade compute, Helm.ai Driver targets production ADAS and higher levels of automation for OEM customers. It works alongside Helm.ai Vision and the company's generative-simulation models so automakers can develop, validate, and ship an end-to-end driving stack, and it underpins Helm.ai's joint development with Honda for next-generation consumer vehicles.
A suite of generative-AI foundation models that synthesize realistic driving video, labeled images, and full driving scenes for training and validating autonomous systems. By generating high-fidelity, diverse, and labeled data on demand, these models let developers stress-test perception and driving policies against rare and dangerous scenarios that are hard to capture in the real world. Generative simulation extends Helm.ai's Deep Teaching philosophy into the data-generation layer, supporting scalable development of ADAS and autonomous-driving software.
We don't have a live feed for this company's ATS. Their careers page has every open role.
View all careers ↗