Co-Founder, Senior AI and Data Science Engineer

Location: United States or Canada Only >> Remote | Full-Time

Industry: AI-Native Decision Intelligence | SaaS | Data Analytics

About the Role

We're looking for a Senior AI and Data Science to own the intelligence layer of SQOR.ai. That means the predictive and causal engines, the agent orchestration that routes every question, the LLM contracts that turn computed math into executive-grade explanations, and the closed-loop learning infrastructure that makes the platform sharper with every interaction.

This is not a research seat. This is the person who lands models, agents, and feedback systems in production and keeps them honest under live client load. Code first, slides last.

Strong LLM and machine learning experience are non-negotiable. Time-series forecasting, causal inference, prompt and agent engineering, retrieval architectures, evaluation harnesses, and the discipline to run AI coding tools without letting them drift. All of it on your fingertips.

You'll work alongside the 18 of us who have bootstrapped SQOR and our founding engineering team. We are now rolling out across concurrent client engagements. Our recent partnership with Google has opened enormous opportunity, and the platform is in active POVs with real customers. You're joining at the inflection point: the math works, the agents work, the products work. You get to make them durable, fast, self-correcting, and trustworthy at scale.

🎥 Watch the reel Google made announcing our partnership: https://youtu.be/cu1YR2dRuUg?si=m0N_Gs59luwCkzN0

About SQOR.ai

SQOR.ai is an AI-native Decision Intelligence platform that represents a fundamental shift in how companies relate to their data. We don't build dashboards.

We don't wrap LLMs around SQL and call it AI.

Our platform separates machine learning from language generation by design. ML computes. The LLM explains. The result is a system where the math is deterministic, hallucination is structurally impossible, and every recommendation carries a confidence score informed by causal analysis, not correlation.

We connect directly to the systems companies already use, automatically extracting and analyzing thousands of KPIs to deliver instant answers, predictions, and recommendations. Our agentic framework shares a unified KPI ontology, a central causation engine, and a contextualization system that routes every query through domain expertise, company structure, and user-level personalization before generating a response.

Two provisional patents filed. Active client engagements generating revenue. Building deeply within the Google Cloud ecosystem. Just named by Gartner in the 2026 Hype Cycle for Data and Machine Learning.

What You'll Do:

Own the predictive and causal engines. Time-series forecasting (VAR/SparseVAR family), causal inference, impulse response analysis, sensitivity testing, goal-gap math, distributional and concentration analytics. You'll harden these so they degrade gracefully on the edge cases that matter in real client data: cross-sectional metrics, sparse history, mostly-constant series. The LLM should never receive garbage to summarize.

Guide the agent orchestration. A routing supervisor decides which specialist agent answers a given question. You'll evolve it so the routing decisions are explainable, evaluable, and self-improving from real interactions. You'll define agent charters that don't overlap, build the evaluation harness that catches routing regressions before they ship, and design the handoffs between ML scoring and generative explanation.

Own the LLM contracts. Prompts versioned and tested. Tool calls with explicit scope. Retrieval architectures built and benchmarked. Prompt-injection defenses in place. Fine-tuning and distillation when they actually pay off. You will treat LLMs as engineering primitives with measurable contracts, not as magic.

Optimize for response time, token utilization, and response quality. Every user interaction has a latency budget, a token budget, and a quality bar. You'll instrument all three, identify the dominant cost drivers (prompt length, retrieval misses, agent fan-out, summarizer thrash), and drive them down without sacrificing the answer. Where models matter, you'll pick smaller-and-faster over larger-and-slower whenever the quality bar permits.

Build the closed-loop learning infrastructure. Every accepted answer, corrected answer, escalation, and silence is data. You'll architect the loops that turn that data into measurable platform improvement, with telemetry, labeled-set construction, and offline evaluation gates. This is one of the things that further differentiates us from the pack.

Run the truth-testing program. End-to-end gates from question to answer: SQL validation, compute correctness, agent routing accuracy, summarizer fidelity. You'll own the harnesses, make them fast, and make them blocking on every release. A regression should be impossible to ship in silence.

Drive and discipline AI coding collaborators. The CEO and the engineering team are heavily using AI coding tools (Claude Code, Codex, Cursor, etc.). These tools produce a lot of output very fast. Without strict discipline they overreach, bundle authorized and unauthorized changes under one label, paraphrase source material then claim alignment with the original, and frame their own past work as ambient drift. You catch every one of those patterns in real time, demand per-record receipts, and redirect before damage is done. Th

Deliver for clients. Hands-on work with live client data. You'll diagnose, why a query drifts, why a routing call surprises someone, why latency spikes. Calmly, fast, with receipts. You'll work directly with client CIOs, data leaders, and PE deal teams when the situation calls for it.

Mentor and raise the bar. Set the standard for code quality, evaluation rigor, and delivery predictability across the AI and data team. Keep the team unblocked. Create an environment where engineers ship good work on committed timelines without burning out.

Required Skills and Experience:

• 8+ years in machine learning, data science, or AI engineering, with at least 3 in a senior or leadership role
• Production LLM engineering: agent design, tool calling, retrieval, evaluation harnesses, prompt regression testing, prompt-injection defenses. Real systems where LLMs are part of the critical path
• Demonstrated optimization of response time, token utilization, and response quality in production LLM systems. You can describe specific wins on latency, cost-per-call, and answer fidelity
• Time-series forecasting at scale: VAR, SparseVAR, structural breaks, ARIMA family. You know when each applies and when it's the wrong tool
• Causal inference: do-calculus, instrumental variables, propensity scores, sensitivity analysis. Working knowledge of when correlation is not enough and how to prove it
• Advanced Python: FastAPI, async, pandas, sklearn, statsmodels, numpy. You write code other engineers can read and maintain
• Deep SQL fluency on BigQuery: schema awareness, query optimization, cost management, large-scale loading
• Client-facing experience required. You will be in calls with CIOs, data leaders, and PE deal teams. You can explain technical decisions to non-technical audiences without dumbing them down, and you can handle a hot room
• Demonstrated track record disciplining AI coding tools: you can show specific examples of catching tool drift, redirecting it, and integrating only the correct output

Preferred Qualifications

• Experience operating ML in Google Cloud / Vertex AI in production
• Open-source contributions or public writing on AI engineering, ML systems, or applied causal inference
• Mid-market ERP data context: complex schemas, multi-entity structures, transactional and reporting layers
• Working knowledge of NL2SQL systems and the failure modes that come with them
• Experience with sparse models, distillation, quantization, and on-the-edge inference
• PE portfolio company or management consulting background where you delivered against external pressure

Why Join Us?

• Own the intelligence layer of one of the most technically substantial Decision Intelligence platforms being built today
• Work on problems that do not exist in traditional BI: deterministic math + LLM explanation separation, agent orchestration with real evaluation, causal scoring on live transactional data, closed-loop learning infrastructure
• Join a fast-moving, remote-first founding-era team where your work has direct, visible impact on every client engagement
• Equity-heavy compensation with real IP behind it: two provisional patents, active revenue, and a capital raise planned

Be part of a paradigm shift, not an incremental improvement