Mohammed Kerraz

• Owned the AI and technical roadmap, translating business priorities into product decisions and production-ready solutions. • Built and evolved an AI-native e-commerce platform where users can discover, compare, and purchase products through a conversational interface. • Designed AI systems using LLMs, agents, RAG, vector search, and ranking, with a strong focus on reliability, scalability, latency, and cost. • Worked across product, business, and engineering, from identifying opportunities and defining requirements to implementation, deployment, and continuous improvement.

• Designed and deployed a production grade financial analysis platform powered by language models and a multi agent architecture, with specialized agents for macroeconomic analysis, market commentary generation, and risk assessment • Developed end to end data pipelines aggregating Bloomberg, FRED, and third-party financial feeds, with data quality controls and transformations feeding the AI models, plus custom macroeconomic indicators used in automated analyses • Built interactive dashboards enabling portfolio managers to consume and act on automated insights, ensuring reliability and observability of the platform in production • Led a team of three engineers across AI and data workstreams, owned the product roadmap and prioritization, and collaborated directly with portfolio managers to translate business needs into concrete features, demonstrating strong stakeholder communication and technical leadership

• Designed and deployed a LangGraph multi agent pipeline orchestrating multi step reasoning, tool calls, and synthesis to generate VC research memos grounded in verifiable sources • Built the full RAG layer end to end: ingestion of web and document snippets, chunking, embedding generation, retrieval, and re ranking of results before synthesis to maximize answer quality • Established a golden evaluation set to regression test prompt and tool changes across runs, bringing systematic evaluation and governance practices to LLM development • Led discovery sessions with more than 20 clients and aligned product features with real investment workflows, showing strong client empathy, active listening, and the ability to translate ambiguous business needs into shipped product

• Facilitated workshops with traders and business teams to formalize needs, analyzed operational processes, and identified high value automation and AI use cases, showing strong active listening and the ability to bridge business and technical audiences • Built decision support prototypes and proofs of concept using Python, LLMs, and analytics tooling, and presented results and recommendations to business stakeholders with clear, persuasive communication • Benchmarked more than 15 specialized AI and analytics vendors and contributed to the selection of two solutions ultimately acquired by the company, combining analytical rigor with sound business judgment

Prepared and analyzed over 10 years of financial and energy time series in Python and built experimentation sets for rigorous evaluation of forecasting models • Improved LLM based forecasting by developing a prompting method that injects technical market signals, achieving roughly a 10% gain over the TimeGPT baseline • Designed an algorithm for automatic detection of support and resistance levels, demonstrating autonomy, scientific rigor, and the ability to communicate findings in an international research environment