• Building and deploying AI agents and agentic workflows using the client's internal agent framework and modern orchestration tools (LangChain, LangGraph, CrewAI, AutoGen, or similar) • Designing and implementing RAG pipelines - including chunking strategies, embeddings, and vector store integrations (Pinecone, Weaviate, pgvector) • Processing and structuring financial documents (PDFs, DOCX reports, tables, CIM/DDQ materials) into clean, machine-readable outputs via Python • Integrating REST APIs and cloud services to connect agent workflows with existing business systems and data infrastructure • Owning the full delivery cycle for each initiative: scoping, development, testing, deployment, and handover to business users • Instrumenting agents for observability, writing test harnesses for non-deterministic behaviour, and ensuring failures are explicit and handled gracefully • Communicating proactively with the client-side team - surfacing blockers early, documenting solutions clearly, and keeping stakeholders in the loop without being asked

• Serve as the technical liaison between Nsight and the IVA provider • Manage day-to-day performance against quality and HIPAA compliance standards • Define acceptance criteria and build layered test harnesses for all provider releases • Configure and tune AI phone agents to meet clinical and operational goals • Build and own the alerting layer • Architect AI-driven quality intelligence pipelines • Build PHI-safe audio processing pipelines • Build the tooling the work requires

• Partner with business and technical stakeholders to understand workflows, challenges, and success metrics • Translate ambiguous problems into clear technical solutions, designs, and delivery plans • Design and build AI-powered applications using LLMs, APIs, and enterprise data systems • Develop production-grade backend services using Python and frameworks like FastAPI or Flask • Build and maintain RAG systems, including document ingestion and normalization, chunking, metadata, and embedding strategies, vector, keyword, and hybrid search, reranking and relevance tuning, source attribution and grounding • Integrate AI solutions with SharePoint, Microsoft Graph, SQL databases, internal APIs, and business applications • Design secure systems that respect access control, governance, and enterprise compliance requirements • Build observable, reliable, and maintainable AI workflows in production environments • Establish evaluation frameworks for LLM systems (accuracy, groundedness, latency, completeness, and failure modes) • Iterate rapidly through prototypes, pilots, and production releases based on user feedback • Collaborate with engineering, data, security, and product teams to ensure adoption and long-term sustainability • Make pragmatic technology choices across Azure OpenAI, Azure AI Search, and open-source tools • Document architecture, decisions, trade-offs, and operational requirements.

• Design, build, and optimize ML/DL models for production-scale audio deepfake detection, ensuring robustness across diverse real-world conditions including compression artifacts, noise, telephony, and streaming pipelines. • Partner with clients to develop a deep understanding of their production environments and define model performance criteria. • Investigate failure cases in client environments, build custom evaluation frameworks, and implement mitigation strategies spanning both Engineering and AI. • Design and execute structured experimentation roadmaps aligned with client requirements and proactive system resilience goals. Translate findings into clear and actionable insights. • Monitor, measure, and report on model performance in production using data analytics and AI observability tools (e.g. Datadog, Metabase). Identify degradation trends, data drift, and emerging threat patterns before they impact client outcomes. • Build and maintain dashboards and analytics pipelines that surface model health metrics, enabling data-driven decisions across AI, Engineering, and Product teams. • Collaborate with cross-functional partners — Applied Scientists, Deployment Engineers, and Product teams — to deploy scalable, production-grade models with clear performance benchmarks and monitoring in place.