16 Nov
|
SiriusAI
|
Gurugram
About Sirius AI Sirius AI is an AI-first consulting and solutions firm serving financial services clients: banks, fintechs, payments companies and wealth managers. We help enterprises move from AI ideas to production-ready solutions using GenAI, agentic AI, ML platforms and advanced analytics, delivering measurable business impact. Our strength lies in combining FS domain expertise, hands-on AI delivery and enterprise-grade execution. Role Overview We are looking for an Associate Director – AI Foundation Model Delivery to lead the delivery of foundation-model-based solutions for our financial services clients in the US. You will own how transformers-based models are selected, adapted, evaluated, deployed and governed in production, on time, on budget and to regulated-industry standards. This role sits at the intersection of applied research and enterprise delivery. You should be equally comfortable reviewing a fine-tuning experiment, debating model choice and inference cost with a client architect, and walking a model risk team through an evaluation pack. You will lead a team of AI and ML engineers, many of them early in their careers, and set the technical bar for how Sirius AI works with foundation models.
Key Responsibilities
• Own end-to-end delivery of foundation model programs, from use-case framing and model selection through data preparation, adaptation, evaluation, deployment and ongoing monitoring.
• Lead model selection and build-vs-buy decisions across commercial and open-weight models, balancing accuracy, latency, cost, data residency and regulatory constraints.
• Direct model adaptation work including prompt engineering, RAG, supervised fine-tuning, parameter-efficient methods (LoRA/PEFT), instruction tuning and preference optimisation (e.g. DPO, RLHF).
• Oversee training and fine-tuning data pipelines: sourcing, curation, labelling, synthetic data generation, quality assurance and lineage, with strict handling of sensitive FS data.
• Build rigorous evaluation frameworks that combine automated benchmarks, domain-specific test sets, human review, hallucination and bias testing, safety and red-team evaluations.
• Take models into production with sound AIOps: versioning, reproducibility, deployment pipelines, inference optimisation (quantisation, distillation, batching, caching) and cost controls.
• Ensure every model meets enterprise AI governance, model risk management, security and privacy requirements, with audit-ready model documentation and validation evidence.
• Own delivery economics for foundation model work, including scope, estimates, GPU and compute budgets, staffing plans, utilisation and margin.
• Lead, mentor and grow a team of AI/ML engineers and data scientists; build training paths in foundation model techniques and FS domain knowledge.
• Build reusable assets such as fine-tuning recipes, evaluation harnesses, model cards and reference architectures that speed up delivery across engagements.
• Act as senior technical lead with client stakeholders and support pre-sales with solution design, model strategy, estimation and staffing plans. What We're Looking For
• Bachelor's or master's degree in computer science, Machine Learning, Statistics,
Mathematics or a related field; a PhD is a plus.
• 8+ years of experience in AI/ML, data science or ML engineering, including at least 3–4 years working hands-on with LLMs or other foundation models.
• A proven track record of delivering foundation-model-based solutions into production in enterprise environments, with specific examples of fine-tuning, evaluation and deployment.
• Deep technical knowledge of transformer architectures, fine-tuning and alignment methods, RAG, model evaluation and inference optimisation.
• Robust Python skills and hands-on experience with frameworks such as PyTorch, Hugging Face Transformers, PEFT, vLLM/TGI and distributed training (e.g. CUDF, DASK, DeepSpeed, FSDP).
• Experience with cloud AI and GPU platforms (AWS SageMaker/Bedrock, Azure ML/Azure OpenAI, Google Vertex AI, Databricks) and AIOps tooling.
• Experience leading technical delivery teams in a consulting, AI services or product engineering environment, including developing junior talent.
• Ability to translate model trade-offs, risks and evaluation results into clear business and risk language for senior stakeholders. Good to Have
• Delivery experience with financial services clients in areas such as risk, fraud, compliance, credit, customer servicing or research.
• Familiarity with model risk management, model validation practices and responsible AI frameworks.
• Experience with multimodal models, document understanding, speech or small language models for on-premise deployment.
• Exposure to agentic AI, LLM orchestration and tool-use patterns built on top of foundation models.
• Publications, open-source contributions or conference talks in LLMs, Foundation Model or applied ML.
📌 Associate Director (Gurugram)
🏢 SiriusAI
📍 Gurugram