Product · AI systems · Enterprise workflows
Building AI-powered enterprise products for complex operational workflows.
Six years across enterprise SaaS, AI systems, and operational tooling — turning complex backend-heavy products into workflows people actually want to use. I sit between engineering and the business, translate technical depth into clear product decisions, and ship AI features that hold up in production.
About
I work on enterprise products where the interesting problems live underneath the interface — in data flows, integrations, retrieval quality, and the operational rituals that surround a platform.
I trained as an engineer at CentraleSupélec and spent my early career in applied NLP research before moving into product. That background still shapes how I work: I read the API before the mockup, I argue about retrieval quality and observability before I argue about UI polish, and I’m comfortable in the room where engineering tradeoffs get decided.
For the past three years I’ve led product for a B2B SaaS platform used by 70k+ end users across 50+ enterprise customers in 17 countries — owning the roadmap, AI initiatives, RFP solutioning, and the slow work of making a complex operational tool feel obvious to use. What I care about most: translating technical depth into product decisions a non-technical stakeholder can stand behind, and shipping AI features that are useful, evaluated, and reliable in production — not demos.
I think in dataflows, integration boundaries, and where state lives. Most product problems on enterprise platforms sit upstream of the UI — that’s where I start scoping.
Workflows compose, integrations come early, and what we ship to one customer needs to scale to fifty. I scope features around contracts, not screens.
Evaluation, retrieval quality, and human-in-the-loop fallbacks come before the prompt. I treat LLM features like infrastructure, with the same bar for reliability.
Observability, written decisions, and tight support → sprint loops are how complex products stay legible — to the team building them and the people running them.
User interviews, support-ticket mining, telemetry, and shadow sessions before scope is fixed. I’d rather lose a sprint to understanding than ship the wrong thing well.
The schema, the API surface, and the integration map before the PRD. Constraints written down early kill a lot of bad ideas cheaply.
Sales, CS, engineering, and the exec room are reading different surfaces of the same problem. My job is to keep the underlying story coherent across all of them.
PRDs, decision logs, retros, post-mortems — written, dated, linkable. The doc is the spec, the audit trail, and the onboarding ramp for the next person on the team.
Golden sets, offline retrieval metrics, regression suites, and a clear fallback when the model is wrong. The interesting AI work happens after the demo.
Grafana, Sentry, structured logs, support → sprint feedback loops. If a behaviour isn’t instrumented, we’re guessing — and I’d rather not.
Experience
Leading product strategy, AI initiatives, and cross-functional delivery for enterprise SaaS products used in large-scale operational environments. Focused on simplifying complex workflows, scaling platform capabilities, and integrating AI into real operational processes.
Shaped functional evolutions for a large-scale platform inside a highly regulated environment — translating complex legal and technical requirements into product specifications that engineering and institutional stakeholders could both agree on.
An applied-research chapter at the intersection of NLP and product — the place I first fell in love with making models genuinely useful inside a real product surface.
Education & Expertise
A systems-thinking foundation that still shapes how I scope problems, design platforms, and reason about complex data-intensive products today.
Three years of core engineering: maths, signal, software, and the habits of mind that make the rest of the work easier.
Foundational PO certification: product backlog stewardship, value-driven prioritization, and the Scrum framework in practice.
Cross-functional leadership, stakeholder management, customer adoption, enterprise integrations, roadmap execution.
LLMs, RAG, semantic retrieval, prompt engineering, NLP, AI workflow automation, contextual ranking, A/B testing.
SQL, REST APIs, OpenAI APIs, Elasticsearch, AWS, Azure, Grafana, Kibana, Sentry, Jira, Confluence.
Practitioner Notes
What started as a simple AI design experiment slowly became part of my day-to-day PM workflow. Notes on design systems, token burn, deck generation, frontend analysis, and where Claude Design genuinely changes product velocity.
Oumaima Chahid · PM practitioner notes · May 2026 Building a golden set from real support tickets, scoring retrieval offline, and the dashboards we now refuse to ship without.
Why the contract beats the screen on enterprise SaaS, and how scoping around endpoints changed our roadmap shape.
How tightening the path from Sentry alert to Jira ticket to backlog cut reported product issues 20% in two quarters.
A framework for partnering with Sales on architecture, pricing, and delivery without committing to features that will haunt the roadmap.
Workshops, internal tooling, and the small wins that built credibility before we asked anyone to trust an LLM with their workflow.