If you want to move beyond evaluation harnesses and fine-tuning wrappers and do the most consequential, foundational work of your career, let’s talk. OPPORTUNITY SUMMARY - Role: Founding AI Engineer: one of our first engineers, solving what LLMs cannot in high-stakes settings with neuro-symbolic foundation models that are glass-box, aligned and hallucination-free. - Company: We are a well-funded seed-stage AI startup in the Bay Area, California, backed by top-tier VCs. - Where: San Jose, California, in person preferred. Hybrid work, a base outside the US, and visa support (O-1A, H-1B transfer, TN) are also possible. - Pay: Top of the market for a seed-stage company: base salary of $200K–$400K (adjusted for location elsewhere), plus 0.5–2% founding-engineer equity. - Apply: Your resume, your Google Scholar profile if you have one, and one thing you built. No LeetCode-style interviews. _______________________________________________ ABOUT US The next frontier of AI will not be won by prompting ever-larger black boxes or building ever more complex harnesses. The most valuable uses of AI, in healthcare, finance and law, cannot tolerate hallucinations, opacity or unverified guesses; they demand verifiable truth, full auditability and reliably aligned behavior. We are building a neuro-symbolic foundation model for these high-stakes uses: glass-box, verifiably correct at scale, and without sacrificing frontier performance. Our first deployments are in healthcare with active enterprise design partners, with finance and law to follow. Our founding team includes IIT Delhi alumni and PhDs from Princeton and Imperial College London, with deep experience shipping AI systems at Apple and Meta. THE ROLE You will be one of our first engineers and part of the founding engineering team, tackling the root architectural reasons conventional LLMs fail in mission-critical applications. Working directly with the founders, you will own major workstreams end to end, review what ships, and advise closely on the roadmap the founders set. In your first year you will take a model from a reproducible baseline to a design partner’s live evaluation, and help hire the next engineers. The role offers exceptional equity upside and room to grow into technical leadership. WHAT YOU WILL WORK ON - Architecture and model design. Research, design and implement novel neural architectures (Transformers and beyond) in PyTorch from scratch; tune capacity, depth, width and memory footprint against accuracy, latency and hardware limits. - High-stakes reasoning and alignment. Models that extract truth from massive unstructured data, reason transparently, and build in reliability through constrained decoding, inference-time search, and abstaining when the evidence is insufficient. - Frontier training. Rigorous pipelines for pre-training, fine-tuning and reinforcement learning on multi-node GPU clusters: hyperparameter sweeps, ablations and validation against domain-specific benchmarks. - Evaluation, data and serving. Hunting down hallucinations with judge-model evaluation, blinded holdouts and regression gates; synthetic data verified by model critics; ultra-low-latency inference and lookup over structured knowledge at the scale of 100M+ entities. - Medicine and other critical domains. Clinical terminologies (UMLS, SNOMEDCT, RxNorm) and integration with design partners’ mission-critical data and evaluation suites. REQUIRED - An MS, MTech or PhD (completed or in progress) in computer science, electrical engineering or a related field, or an equivalent research record. - Research experience designing, training and evaluating neural models, demonstrated through papers, a thesis or notable open-source work. - You have personally trained and served models end to end, with depth in either model or systems work. PREFERRED - A PhD in machine learning, NLP or a related computational field. - Experience with knowledge graphs, information extraction, constrained or structured decoding, reinforcement learning for large models, inference engines (vLLM, SGLang) or distributed training(FSDP, DeepSpeed). - Biomedical data experience (UMLS, SNOMED CT, RxNorm) or partner-facing enterprise deployments. Nobody has all of these. The strongest signal: you have shipped a model whose outputs were verified against a formal schema, a verifier or a graph rather than left to chance. ATTITUDE MATTERS AS MUCH AS SKILL - You think big. We have the compute and the resources to bring your most ambitious ideas to life, and we want people who will push the frontier. - You own outcomes, not tickets. Ambiguity is where you do your best work, and you treat the gaps between model ideas and customer reality as yours. - Evidence first. Every result ships with its baseline, its experiment and reproducible numbers, and you change your mind when the evidence does. - Truth over herd mentality. You believe AI must be aligned, auditable and verifiably reliable by design to solve high-value problems. “Pacing the frontier” is not a viable plan: the foundation has to be fixed from the ground up. Probably not a fit if you want a narrow lane, prefer managing to building, or want to build prompt-chaining wrappers. HOW WE HIRE No LeetCode-style interviews: we assume you build with tools such as Cursor, Claude Code or Codex, because we do. Instead: a 30-minute founder conversation; a 90-minute deep dive on code you wrote and a failure you debugged; a paid two-to-three-day work sample, done with your usual tools; and references. WHAT WE OFFER - A place on the founding engineering team, with real influence on architecture and research culture. - Hard, unsolved problems at the frontier of AI, the compute to test ambitious ideas at real scale, and direct access to enterprise design partners in high-stakes domains. - Top-of-market pay, 0.5–2% founding-engineer equity, and visa support if needed (O-1A, H-1B transfer, TN). HOW TO APPLY Submit your application with your resume, your Google Scholar profile if you have one, and one thing you built, with a few lines on what you owned. We are an equal opportunity employer. We welcome applicants of every background and do not discriminate on the basis of race, color, religion, sex, sexual orientation, gender identity, national origin, age, disability, veteran status or any other legally protected characteristic. If you need an accommodation at any point in the process, tell us and we will arrange it.