Director - Data Science

ABOUT INCEDO:

Incedo is a global AI and data transformation specialist, helping companies turn digital investment into sustainable business impact by delivering ROI from AI @ Scale.

We are 4,000+ people across the US, Canada, Latin America and India, working with Fortune 500 enterprises and fast-growing clients in banking & payments, wealth management, telecom, hi-tech and life sciences.

Build what's next in AI, data and enterprise platforms:

Our Platform & Solutions portfolio is where Incedo builds AI-native products for real enterprise problems: Incedo Lighthouse (AI-powered decision intelligence), DataXel (agentic data modernization), brAInspark (agentic AI enablement), IncedoPay (integrated payables), Kratos (regulatory compliance and data control for banking) and DQXpert (AI-powered data quality) - plus domain products across customer support, document processing, quality engineering, healthcare and financial services.

You'll work with multidisciplinary teams across Product, Engineering, AI/Data, Design and Business, taking ideas from problem discovery and experimentation to production-grade platforms that enterprises depend on.

WHY THIS ROLE:

Most Director of Data Science roles sit in one camp: a pure analytics function, or a GenAI product team. This one needs both, working together.

DataXel needs a data science Leader who can design autonomous agents that take consequential actions on live enterprise data estates - safely.

DQXpert needs someone who can make generative AI trustworthy enough that business users will actually act on what it tells them about their data.

THE ROLE:

You will be the Data Science Leader for DataXel and DQXpert, Incedo's data modernization product suite. You set the modelling and agentic architecture strategy, lead the data science bench, and own the intelligence layer of both products from prototype to production.

Most Director of Data Science roles sit either inside a pure analytics function or on top of a GenAI product. This one needs both. DataXel is not a model - it is an autonomous agent taking consequential actions on live enterprise data estates, so it needs someone fluent in multi-step planning, tool orchestration and safe autonomous execution. DQXpert needs generative AI trustworthy enough that business users act on what it tells them - which takes classical ML rigour to build the deterministic guardrails underneath.

WHAT YOU'LL OWN:

1. The agentic modernization engine - DataXel:

- Lead the design and evolution of DataXel's agentic architecture: autonomous schema discovery, intelligent mapping, pipeline generation and migration execution, orchestrated across a multi-step agentic loop.

- Define the human-in-the-loop gate framework - where approval is required, how confidence and uncertainty are surfaced to the reviewer, and how the system escalates gracefully when thresholds aren't met.

- Build the learning system that turns every HITL approval and rejection into a feedback signal that improves future automation.

- Partner with data engineering to ensure the agentic execution layer integrates safely with DataXel's transformation engine.

2. GenAI-powered data quality - DQXpert:

- Lead the design of DQXpert's GenAI layer: natural-language data profiling, LLM-based anomaly characterisation, and plain-language DQ rule authoring that turns business intent into executable validation logic.

- Architect the knowledge context layer behind remediation recommendations - how domain knowledge, past fixes and lineage context get retrieved and injected into prompts to produce specific, trustworthy suggestions instead of generic ones.

- Build the continuous improvement loop: track accept/reject behaviour on rules and recommendations, and use it to sharpen what the system surfaces.

- Own the classical ML and statistical models where deterministic correctness beats generative fluency - anomaly detection baselines, schema drift scoring, and the DQ metrics the GenAI layer reasons on top of.

3. Platform architecture & MLOps:

- Define the AI architecture across the suite: model versioning, agentic workflow orchestration, RAG pipeline management, feedback instrumentation and production monitoring.

- Ensure every autonomous action in DataXel and every generated output in DQXpert has defined confidence thresholds, fallback behaviour and an audit trail - so both products fail safely and visibly.

- Champion responsible AI: hallucination mitigation, PII-safe prompting, compliance-conscious design, and explainability that end users can follow.

4. Team leadership, Client Engagement & Presales:

- Mentoring and growing the data science bench, partnering with Engineering and Product on roadmap, and representing both platforms credibly in client conversations, POCs and pre-sales.

WHAT YOU'LL BRING:

- 14 - 18 years in data science and applied AI, including 3+ years leading technically at Director level or equivalent - and still hands-on.

- Deep classical ML: anomaly detection, time-series, schema matching, entity resolution, clustering, classification - enough depth to build the guardrails that make generative output trustworthy.

- Production LLM/GenAI engineering: prompt design, RAG architectures, fine-tuning, knowledge-context design.

- Hands-on agentic platform design: multi-step LLM orchestration, tool use, autonomous execution and HITL patterns - LangGraph, LangChain, CrewAI or equivalent.

- Feedback-loop systems: RLHF-adjacent methods, behavioural signal capture, continuous improvement from human interaction data.

- Production-grade Python (pandas, scikit-learn, PySpark), plus working fluency with data pipeline and ETL ecosystems (Informatica, dbt, Spark, AWS Glue or equivalent).

- Cloud data platforms: AWS (S3, Redshift, Glue, EMR) or Azure/GCP equivalent, and lakehouse architectures.

- MLOps/LLMOps in production: model registries, automated retraining, drift detection, RAG pipeline monitoring.

- Responsible AI instincts: hallucination mitigation, PII-safe prompting, explainability and audit-ready design.

Nice to have:

- Built data quality, observability or data profiling intelligence - in a product company or at serious scale in client delivery.

- Familiarity with data mesh or data fabric patterns and what they mean for distributed data quality enforcement.

- AWS certifications (Machine Learning Specialty, Data Analytics Specialty or equivalent).

- Consulting or SI background with exposure to enterprise-scale data transformation programmes.

Education:

- B. Tech / M. Tech / M. Sc. in Computer Science, Statistics, Mathematics or a closely related technical discipline. Equivalent industry experience considered.

Skills

Data Science, Data Management, Data Quality, Generative AI, Artificial Intelligence, Machine Learning, Statistics